cm.c 114.2 KB
Newer Older
1
/*
2
 * Copyright (c) 2009-2014 Chelsio, Inc. All rights reserved.
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
 *
 * This software is available to you under a choice of one of two
 * licenses.  You may choose to be licensed under the terms of the GNU
 * General Public License (GPL) Version 2, available from the file
 * COPYING in the main directory of this source tree, or the
 * OpenIB.org BSD license below:
 *
 *     Redistribution and use in source and binary forms, with or
 *     without modification, are permitted provided that the following
 *     conditions are met:
 *
 *      - Redistributions of source code must retain the above
 *	  copyright notice, this list of conditions and the following
 *	  disclaimer.
 *
 *      - Redistributions in binary form must reproduce the above
 *	  copyright notice, this list of conditions and the following
 *	  disclaimer in the documentation and/or other materials
 *	  provided with the distribution.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 * SOFTWARE.
 */
#include <linux/module.h>
#include <linux/list.h>
#include <linux/workqueue.h>
#include <linux/skbuff.h>
#include <linux/timer.h>
#include <linux/notifier.h>
#include <linux/inetdevice.h>
#include <linux/ip.h>
#include <linux/tcp.h>
41
#include <linux/if_vlan.h>
42 43 44 45

#include <net/neighbour.h>
#include <net/netevent.h>
#include <net/route.h>
46
#include <net/tcp.h>
47 48
#include <net/ip6_route.h>
#include <net/addrconf.h>
49

S
Steve Wise 已提交
50 51
#include <rdma/ib_addr.h>

52
#include "iw_cxgb4.h"
H
Hariprasad S 已提交
53
#include "clip_tbl.h"
54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70

static char *states[] = {
	"idle",
	"listen",
	"connecting",
	"mpa_wait_req",
	"mpa_req_sent",
	"mpa_req_rcvd",
	"mpa_rep_sent",
	"fpdu_mode",
	"aborting",
	"closing",
	"moribund",
	"dead",
	NULL,
};

71 72 73 74 75 76 77 78
static int nocong;
module_param(nocong, int, 0644);
MODULE_PARM_DESC(nocong, "Turn of congestion control (default=0)");

static int enable_ecn;
module_param(enable_ecn, int, 0644);
MODULE_PARM_DESC(enable_ecn, "Enable ECN (default=0/disabled)");

S
Steve Wise 已提交
79
static int dack_mode = 1;
80
module_param(dack_mode, int, 0644);
S
Steve Wise 已提交
81
MODULE_PARM_DESC(dack_mode, "Delayed ack mode (default=1)");
82

83
uint c4iw_max_read_depth = 32;
84
module_param(c4iw_max_read_depth, int, 0644);
85 86
MODULE_PARM_DESC(c4iw_max_read_depth,
		 "Per-connection max ORD/IRD (default=32)");
87

88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104
static int enable_tcp_timestamps;
module_param(enable_tcp_timestamps, int, 0644);
MODULE_PARM_DESC(enable_tcp_timestamps, "Enable tcp timestamps (default=0)");

static int enable_tcp_sack;
module_param(enable_tcp_sack, int, 0644);
MODULE_PARM_DESC(enable_tcp_sack, "Enable tcp SACK (default=0)");

static int enable_tcp_window_scaling = 1;
module_param(enable_tcp_window_scaling, int, 0644);
MODULE_PARM_DESC(enable_tcp_window_scaling,
		 "Enable tcp window scaling (default=1)");

int c4iw_debug;
module_param(c4iw_debug, int, 0644);
MODULE_PARM_DESC(c4iw_debug, "Enable debug logging (default=0)");

105
static int peer2peer = 1;
106
module_param(peer2peer, int, 0644);
107
MODULE_PARM_DESC(peer2peer, "Support peer2peer ULPs (default=1)");
108 109 110 111 112 113 114 115 116 117 118

static int p2p_type = FW_RI_INIT_P2PTYPE_READ_REQ;
module_param(p2p_type, int, 0644);
MODULE_PARM_DESC(p2p_type, "RDMAP opcode to use for the RTR message: "
			   "1=RDMA_READ 0=RDMA_WRITE (default 1)");

static int ep_timeout_secs = 60;
module_param(ep_timeout_secs, int, 0644);
MODULE_PARM_DESC(ep_timeout_secs, "CM Endpoint operation timeout "
				   "in seconds (default=60)");

119
static int mpa_rev = 2;
120 121
module_param(mpa_rev, int, 0644);
MODULE_PARM_DESC(mpa_rev, "MPA Revision, 0 supports amso1100, "
122
		"1 is RFC5044 spec compliant, 2 is IETF MPA Peer Connect Draft"
123
		" compliant (default=2)");
124 125 126 127 128 129 130 131 132 133 134 135 136

static int markers_enabled;
module_param(markers_enabled, int, 0644);
MODULE_PARM_DESC(markers_enabled, "Enable MPA MARKERS (default(0)=disabled)");

static int crc_enabled = 1;
module_param(crc_enabled, int, 0644);
MODULE_PARM_DESC(crc_enabled, "Enable MPA CRC (default(1)=enabled)");

static int rcv_win = 256 * 1024;
module_param(rcv_win, int, 0644);
MODULE_PARM_DESC(rcv_win, "TCP receive window in bytes (default=256KB)");

137
static int snd_win = 128 * 1024;
138
module_param(snd_win, int, 0644);
139
MODULE_PARM_DESC(snd_win, "TCP send window in bytes (default=128KB)");
140 141 142 143 144 145 146 147

static struct workqueue_struct *workq;

static struct sk_buff_head rxq;

static struct sk_buff *get_skb(struct sk_buff *skb, int len, gfp_t gfp);
static void ep_timeout(unsigned long arg);
static void connect_reply_upcall(struct c4iw_ep *ep, int status);
148
static int sched(struct c4iw_dev *dev, struct sk_buff *skb);
149

150 151 152
static LIST_HEAD(timeout_list);
static spinlock_t timeout_lock;

153 154 155 156 157 158 159 160 161 162 163 164 165
static void deref_cm_id(struct c4iw_ep_common *epc)
{
	epc->cm_id->rem_ref(epc->cm_id);
	epc->cm_id = NULL;
	set_bit(CM_ID_DEREFED, &epc->history);
}

static void ref_cm_id(struct c4iw_ep_common *epc)
{
	set_bit(CM_ID_REFED, &epc->history);
	epc->cm_id->add_ref(epc->cm_id);
}

166 167 168 169
static void deref_qp(struct c4iw_ep *ep)
{
	c4iw_qp_rem_ref(&ep->com.qp->ibqp);
	clear_bit(QP_REFERENCED, &ep->com.flags);
170
	set_bit(QP_DEREFED, &ep->com.history);
171 172 173 174 175
}

static void ref_qp(struct c4iw_ep *ep)
{
	set_bit(QP_REFERENCED, &ep->com.flags);
176
	set_bit(QP_REFED, &ep->com.history);
177 178 179
	c4iw_qp_add_ref(&ep->com.qp->ibqp);
}

180 181 182 183
static void start_ep_timer(struct c4iw_ep *ep)
{
	PDBG("%s ep %p\n", __func__, ep);
	if (timer_pending(&ep->timer)) {
184 185 186 187 188 189
		pr_err("%s timer already started! ep %p\n",
		       __func__, ep);
		return;
	}
	clear_bit(TIMEOUT, &ep->com.flags);
	c4iw_get_ep(&ep->com);
190 191 192 193 194 195
	ep->timer.expires = jiffies + ep_timeout_secs * HZ;
	ep->timer.data = (unsigned long)ep;
	ep->timer.function = ep_timeout;
	add_timer(&ep->timer);
}

S
Steve Wise 已提交
196
static int stop_ep_timer(struct c4iw_ep *ep)
197
{
198
	PDBG("%s ep %p stopping\n", __func__, ep);
199
	del_timer_sync(&ep->timer);
S
Steve Wise 已提交
200
	if (!test_and_set_bit(TIMEOUT, &ep->com.flags)) {
201
		c4iw_put_ep(&ep->com);
S
Steve Wise 已提交
202 203 204
		return 0;
	}
	return 1;
205 206 207 208 209 210 211 212 213 214 215 216 217 218 219
}

static int c4iw_l2t_send(struct c4iw_rdev *rdev, struct sk_buff *skb,
		  struct l2t_entry *l2e)
{
	int	error = 0;

	if (c4iw_fatal_error(rdev)) {
		kfree_skb(skb);
		PDBG("%s - device in error state - dropping\n", __func__);
		return -EIO;
	}
	error = cxgb4_l2t_send(rdev->lldi.ports[0], skb, l2e);
	if (error < 0)
		kfree_skb(skb);
220
	return error < 0 ? error : 0;
221 222 223 224 225 226 227 228 229 230 231 232 233 234
}

int c4iw_ofld_send(struct c4iw_rdev *rdev, struct sk_buff *skb)
{
	int	error = 0;

	if (c4iw_fatal_error(rdev)) {
		kfree_skb(skb);
		PDBG("%s - device in error state - dropping\n", __func__);
		return -EIO;
	}
	error = cxgb4_ofld_send(rdev->lldi.ports[0], skb);
	if (error < 0)
		kfree_skb(skb);
235
	return error < 0 ? error : 0;
236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254
}

static void release_tid(struct c4iw_rdev *rdev, u32 hwtid, struct sk_buff *skb)
{
	struct cpl_tid_release *req;

	skb = get_skb(skb, sizeof *req, GFP_KERNEL);
	if (!skb)
		return;
	req = (struct cpl_tid_release *) skb_put(skb, sizeof(*req));
	INIT_TP_WR(req, hwtid);
	OPCODE_TID(req) = cpu_to_be32(MK_OPCODE_TID(CPL_TID_RELEASE, hwtid));
	set_wr_txq(skb, CPL_PRIORITY_SETUP, 0);
	c4iw_ofld_send(rdev, skb);
	return;
}

static void set_emss(struct c4iw_ep *ep, u16 opt)
{
255
	ep->emss = ep->com.dev->rdev.lldi.mtus[TCPOPT_MSS_G(opt)] -
256 257 258
		   ((AF_INET == ep->com.remote_addr.ss_family) ?
		    sizeof(struct iphdr) : sizeof(struct ipv6hdr)) -
		   sizeof(struct tcphdr);
259
	ep->mss = ep->emss;
260
	if (TCPOPT_TSTAMP_G(opt))
261
		ep->emss -= round_up(TCPOLEN_TIMESTAMP, 4);
262 263
	if (ep->emss < 128)
		ep->emss = 128;
264 265
	if (ep->emss & 7)
		PDBG("Warning: misaligned mtu idx %u mss %u emss=%u\n",
266 267
		     TCPOPT_MSS_G(opt), ep->mss, ep->emss);
	PDBG("%s mss_idx %u mss %u emss=%u\n", __func__, TCPOPT_MSS_G(opt),
268 269 270 271 272 273 274
	     ep->mss, ep->emss);
}

static enum c4iw_ep_state state_read(struct c4iw_ep_common *epc)
{
	enum c4iw_ep_state state;

275
	mutex_lock(&epc->mutex);
276
	state = epc->state;
277
	mutex_unlock(&epc->mutex);
278 279 280 281 282 283 284 285 286 287
	return state;
}

static void __state_set(struct c4iw_ep_common *epc, enum c4iw_ep_state new)
{
	epc->state = new;
}

static void state_set(struct c4iw_ep_common *epc, enum c4iw_ep_state new)
{
288
	mutex_lock(&epc->mutex);
289 290
	PDBG("%s - %s -> %s\n", __func__, states[epc->state], states[new]);
	__state_set(epc, new);
291
	mutex_unlock(&epc->mutex);
292 293 294 295 296 297 298 299 300 301
	return;
}

static void *alloc_ep(int size, gfp_t gfp)
{
	struct c4iw_ep_common *epc;

	epc = kzalloc(size, gfp);
	if (epc) {
		kref_init(&epc->kref);
302
		mutex_init(&epc->mutex);
303
		c4iw_init_wr_wait(&epc->wr_wait);
304 305 306 307 308 309 310 311 312 313
	}
	PDBG("%s alloc ep %p\n", __func__, epc);
	return epc;
}

void _c4iw_free_ep(struct kref *kref)
{
	struct c4iw_ep *ep;

	ep = container_of(kref, struct c4iw_ep, com.kref);
314
	PDBG("%s ep %p state %s\n", __func__, ep, states[ep->com.state]);
315 316
	if (test_bit(QP_REFERENCED, &ep->com.flags))
		deref_qp(ep);
317
	if (test_bit(RELEASE_RESOURCES, &ep->com.flags)) {
H
Hariprasad S 已提交
318 319 320
		if (ep->com.remote_addr.ss_family == AF_INET6) {
			struct sockaddr_in6 *sin6 =
					(struct sockaddr_in6 *)
321
					&ep->com.local_addr;
H
Hariprasad S 已提交
322 323 324 325 326 327

			cxgb4_clip_release(
					ep->com.dev->rdev.lldi.ports[0],
					(const u32 *)&sin6->sin6_addr.s6_addr,
					1);
		}
328
		remove_handle(ep->com.dev, &ep->com.dev->hwtid_idr, ep->hwtid);
329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373
		cxgb4_remove_tid(ep->com.dev->rdev.lldi.tids, 0, ep->hwtid);
		dst_release(ep->dst);
		cxgb4_l2t_release(ep->l2t);
	}
	kfree(ep);
}

static void release_ep_resources(struct c4iw_ep *ep)
{
	set_bit(RELEASE_RESOURCES, &ep->com.flags);
	c4iw_put_ep(&ep->com);
}

static int status2errno(int status)
{
	switch (status) {
	case CPL_ERR_NONE:
		return 0;
	case CPL_ERR_CONN_RESET:
		return -ECONNRESET;
	case CPL_ERR_ARP_MISS:
		return -EHOSTUNREACH;
	case CPL_ERR_CONN_TIMEDOUT:
		return -ETIMEDOUT;
	case CPL_ERR_TCAM_FULL:
		return -ENOMEM;
	case CPL_ERR_CONN_EXIST:
		return -EADDRINUSE;
	default:
		return -EIO;
	}
}

/*
 * Try and reuse skbs already allocated...
 */
static struct sk_buff *get_skb(struct sk_buff *skb, int len, gfp_t gfp)
{
	if (skb && !skb_is_nonlinear(skb) && !skb_cloned(skb)) {
		skb_trim(skb, 0);
		skb_get(skb);
		skb_reset_transport_header(skb);
	} else {
		skb = alloc_skb(len, gfp);
	}
374
	t4_set_arp_err_handler(skb, NULL, NULL);
375 376 377
	return skb;
}

378 379
static struct net_device *get_real_dev(struct net_device *egress_dev)
{
S
Steve Wise 已提交
380
	return rdma_vlan_dev_real_dev(egress_dev) ? : egress_dev;
381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423
}

static int our_interface(struct c4iw_dev *dev, struct net_device *egress_dev)
{
	int i;

	egress_dev = get_real_dev(egress_dev);
	for (i = 0; i < dev->rdev.lldi.nports; i++)
		if (dev->rdev.lldi.ports[i] == egress_dev)
			return 1;
	return 0;
}

static struct dst_entry *find_route6(struct c4iw_dev *dev, __u8 *local_ip,
				     __u8 *peer_ip, __be16 local_port,
				     __be16 peer_port, u8 tos,
				     __u32 sin6_scope_id)
{
	struct dst_entry *dst = NULL;

	if (IS_ENABLED(CONFIG_IPV6)) {
		struct flowi6 fl6;

		memset(&fl6, 0, sizeof(fl6));
		memcpy(&fl6.daddr, peer_ip, 16);
		memcpy(&fl6.saddr, local_ip, 16);
		if (ipv6_addr_type(&fl6.daddr) & IPV6_ADDR_LINKLOCAL)
			fl6.flowi6_oif = sin6_scope_id;
		dst = ip6_route_output(&init_net, NULL, &fl6);
		if (!dst)
			goto out;
		if (!our_interface(dev, ip6_dst_idev(dst)->dev) &&
		    !(ip6_dst_idev(dst)->dev->flags & IFF_LOOPBACK)) {
			dst_release(dst);
			dst = NULL;
		}
	}

out:
	return dst;
}

static struct dst_entry *find_route(struct c4iw_dev *dev, __be32 local_ip,
424 425 426 427
				 __be32 peer_ip, __be16 local_port,
				 __be16 peer_port, u8 tos)
{
	struct rtable *rt;
428
	struct flowi4 fl4;
429
	struct neighbour *n;
430

431
	rt = ip_route_output_ports(&init_net, &fl4, NULL, peer_ip, local_ip,
432 433
				   peer_port, local_port, IPPROTO_TCP,
				   tos, 0);
434
	if (IS_ERR(rt))
435
		return NULL;
436 437 438
	n = dst_neigh_lookup(&rt->dst, &peer_ip);
	if (!n)
		return NULL;
439 440
	if (!our_interface(dev, n->dev) &&
	    !(n->dev->flags & IFF_LOOPBACK)) {
441
		neigh_release(n);
442 443 444 445 446
		dst_release(&rt->dst);
		return NULL;
	}
	neigh_release(n);
	return &rt->dst;
447 448 449 450
}

static void arp_failure_discard(void *handle, struct sk_buff *skb)
{
451
	pr_err(MOD "ARP failure\n");
452 453 454
	kfree_skb(skb);
}

455
enum {
456
	NUM_FAKE_CPLS = 2,
457
	FAKE_CPL_PUT_EP_SAFE = NUM_CPL_CMDS + 0,
458
	FAKE_CPL_PASS_PUT_EP_SAFE = NUM_CPL_CMDS + 1,
459 460 461 462 463 464 465 466 467 468 469
};

static int _put_ep_safe(struct c4iw_dev *dev, struct sk_buff *skb)
{
	struct c4iw_ep *ep;

	ep = *((struct c4iw_ep **)(skb->cb + 2 * sizeof(void *)));
	release_ep_resources(ep);
	return 0;
}

470 471 472 473 474 475 476 477 478 479
static int _put_pass_ep_safe(struct c4iw_dev *dev, struct sk_buff *skb)
{
	struct c4iw_ep *ep;

	ep = *((struct c4iw_ep **)(skb->cb + 2 * sizeof(void *)));
	c4iw_put_ep(&ep->parent_ep->com);
	release_ep_resources(ep);
	return 0;
}

480 481 482 483 484 485
/*
 * Fake up a special CPL opcode and call sched() so process_work() will call
 * _put_ep_safe() in a safe context to free the ep resources.  This is needed
 * because ARP error handlers are called in an ATOMIC context, and
 * _c4iw_free_ep() needs to block.
 */
486 487
static void queue_arp_failure_cpl(struct c4iw_ep *ep, struct sk_buff *skb,
				  int cpl)
488 489 490 491
{
	struct cpl_act_establish *rpl = cplhdr(skb);

	/* Set our special ARP_FAILURE opcode */
492
	rpl->ot.opcode = cpl;
493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510

	/*
	 * Save ep in the skb->cb area, after where sched() will save the dev
	 * ptr.
	 */
	*((struct c4iw_ep **)(skb->cb + 2 * sizeof(void *))) = ep;
	sched(ep->com.dev, skb);
}

/* Handle an ARP failure for an accept */
static void pass_accept_rpl_arp_failure(void *handle, struct sk_buff *skb)
{
	struct c4iw_ep *ep = handle;

	pr_err(MOD "ARP failure during accept - tid %u -dropping connection\n",
	       ep->hwtid);

	__state_set(&ep->com, DEAD);
511
	queue_arp_failure_cpl(ep, skb, FAKE_CPL_PASS_PUT_EP_SAFE);
512 513
}

514 515 516 517 518
/*
 * Handle an ARP failure for an active open.
 */
static void act_open_req_arp_failure(void *handle, struct sk_buff *skb)
{
519 520
	struct c4iw_ep *ep = handle;

M
Masanari Iida 已提交
521
	printk(KERN_ERR MOD "ARP failure during connect\n");
522
	connect_reply_upcall(ep, -EHOSTUNREACH);
523
	__state_set(&ep->com, DEAD);
H
Hariprasad S 已提交
524 525
	if (ep->com.remote_addr.ss_family == AF_INET6) {
		struct sockaddr_in6 *sin6 =
526
			(struct sockaddr_in6 *)&ep->com.local_addr;
H
Hariprasad S 已提交
527 528 529
		cxgb4_clip_release(ep->com.dev->rdev.lldi.ports[0],
				   (const u32 *)&sin6->sin6_addr.s6_addr, 1);
	}
530 531
	remove_handle(ep->com.dev, &ep->com.dev->atid_idr, ep->atid);
	cxgb4_free_atid(ep->com.dev->rdev.lldi.tids, ep->atid);
532
	queue_arp_failure_cpl(ep, skb, FAKE_CPL_PUT_EP_SAFE);
533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548
}

/*
 * Handle an ARP failure for a CPL_ABORT_REQ.  Change it into a no RST variant
 * and send it along.
 */
static void abort_arp_failure(void *handle, struct sk_buff *skb)
{
	struct c4iw_rdev *rdev = handle;
	struct cpl_abort_req *req = cplhdr(skb);

	PDBG("%s rdev %p\n", __func__, rdev);
	req->cmd = CPL_ABORT_NO_RST;
	c4iw_ofld_send(rdev, skb);
}

549
static int send_flowc(struct c4iw_ep *ep, struct sk_buff *skb)
550 551 552 553
{
	unsigned int flowclen = 80;
	struct fw_flowc_wr *flowc;
	int i;
H
Hariprasad S 已提交
554 555 556 557 558 559 560
	u16 vlan = ep->l2t->vlan;
	int nparams;

	if (vlan == CPL_L2T_VLAN_NONE)
		nparams = 8;
	else
		nparams = 9;
561 562 563 564

	skb = get_skb(skb, flowclen, GFP_KERNEL);
	flowc = (struct fw_flowc_wr *)__skb_put(skb, flowclen);

565
	flowc->op_to_nparams = cpu_to_be32(FW_WR_OP_V(FW_FLOWC_WR) |
H
Hariprasad S 已提交
566
					   FW_FLOWC_WR_NPARAMS_V(nparams));
567 568
	flowc->flowid_len16 = cpu_to_be32(FW_WR_LEN16_V(DIV_ROUND_UP(flowclen,
					  16)) | FW_WR_FLOWID_V(ep->hwtid));
569 570

	flowc->mnemval[0].mnemonic = FW_FLOWC_MNEM_PFNVFN;
571
	flowc->mnemval[0].val = cpu_to_be32(FW_PFVF_CMD_PFN_V
572
					    (ep->com.dev->rdev.lldi.pf));
573 574 575 576 577 578 579 580 581 582 583
	flowc->mnemval[1].mnemonic = FW_FLOWC_MNEM_CH;
	flowc->mnemval[1].val = cpu_to_be32(ep->tx_chan);
	flowc->mnemval[2].mnemonic = FW_FLOWC_MNEM_PORT;
	flowc->mnemval[2].val = cpu_to_be32(ep->tx_chan);
	flowc->mnemval[3].mnemonic = FW_FLOWC_MNEM_IQID;
	flowc->mnemval[3].val = cpu_to_be32(ep->rss_qid);
	flowc->mnemval[4].mnemonic = FW_FLOWC_MNEM_SNDNXT;
	flowc->mnemval[4].val = cpu_to_be32(ep->snd_seq);
	flowc->mnemval[5].mnemonic = FW_FLOWC_MNEM_RCVNXT;
	flowc->mnemval[5].val = cpu_to_be32(ep->rcv_seq);
	flowc->mnemval[6].mnemonic = FW_FLOWC_MNEM_SNDBUF;
584
	flowc->mnemval[6].val = cpu_to_be32(ep->snd_win);
585 586
	flowc->mnemval[7].mnemonic = FW_FLOWC_MNEM_MSS;
	flowc->mnemval[7].val = cpu_to_be32(ep->emss);
H
Hariprasad S 已提交
587 588 589 590 591 592 593 594 595 596 597
	if (nparams == 9) {
		u16 pri;

		pri = (vlan & VLAN_PRIO_MASK) >> VLAN_PRIO_SHIFT;
		flowc->mnemval[8].mnemonic = FW_FLOWC_MNEM_SCHEDCLASS;
		flowc->mnemval[8].val = cpu_to_be32(pri);
	} else {
		/* Pad WR to 16 byte boundary */
		flowc->mnemval[8].mnemonic = 0;
		flowc->mnemval[8].val = 0;
	}
598 599 600 601 602 603 604
	for (i = 0; i < 9; i++) {
		flowc->mnemval[i].r4[0] = 0;
		flowc->mnemval[i].r4[1] = 0;
		flowc->mnemval[i].r4[2] = 0;
	}

	set_wr_txq(skb, CPL_PRIORITY_DATA, ep->txq_idx);
605
	return c4iw_ofld_send(&ep->com.dev->rdev, skb);
606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651
}

static int send_halfclose(struct c4iw_ep *ep, gfp_t gfp)
{
	struct cpl_close_con_req *req;
	struct sk_buff *skb;
	int wrlen = roundup(sizeof *req, 16);

	PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid);
	skb = get_skb(NULL, wrlen, gfp);
	if (!skb) {
		printk(KERN_ERR MOD "%s - failed to alloc skb\n", __func__);
		return -ENOMEM;
	}
	set_wr_txq(skb, CPL_PRIORITY_DATA, ep->txq_idx);
	t4_set_arp_err_handler(skb, NULL, arp_failure_discard);
	req = (struct cpl_close_con_req *) skb_put(skb, wrlen);
	memset(req, 0, wrlen);
	INIT_TP_WR(req, ep->hwtid);
	OPCODE_TID(req) = cpu_to_be32(MK_OPCODE_TID(CPL_CLOSE_CON_REQ,
						    ep->hwtid));
	return c4iw_l2t_send(&ep->com.dev->rdev, skb, ep->l2t);
}

static int send_abort(struct c4iw_ep *ep, struct sk_buff *skb, gfp_t gfp)
{
	struct cpl_abort_req *req;
	int wrlen = roundup(sizeof *req, 16);

	PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid);
	skb = get_skb(skb, wrlen, gfp);
	if (!skb) {
		printk(KERN_ERR MOD "%s - failed to alloc skb.\n",
		       __func__);
		return -ENOMEM;
	}
	set_wr_txq(skb, CPL_PRIORITY_DATA, ep->txq_idx);
	t4_set_arp_err_handler(skb, &ep->com.dev->rdev, abort_arp_failure);
	req = (struct cpl_abort_req *) skb_put(skb, wrlen);
	memset(req, 0, wrlen);
	INIT_TP_WR(req, ep->hwtid);
	OPCODE_TID(req) = cpu_to_be32(MK_OPCODE_TID(CPL_ABORT_REQ, ep->hwtid));
	req->cmd = CPL_ABORT_SEND_RST;
	return c4iw_l2t_send(&ep->com.dev->rdev, skb, ep->l2t);
}

652
static void best_mtu(const unsigned short *mtus, unsigned short mtu,
653
		     unsigned int *idx, int use_ts, int ipv6)
654
{
655 656 657
	unsigned short hdr_size = (ipv6 ?
				   sizeof(struct ipv6hdr) :
				   sizeof(struct iphdr)) +
658
				  sizeof(struct tcphdr) +
659 660
				  (use_ts ?
				   round_up(TCPOLEN_TIMESTAMP, 4) : 0);
661 662 663 664 665
	unsigned short data_size = mtu - hdr_size;

	cxgb4_best_aligned_mtu(mtus, hdr_size, data_size, 8, idx);
}

666 667
static int send_connect(struct c4iw_ep *ep)
{
668 669 670 671 672 673
	struct cpl_act_open_req *req = NULL;
	struct cpl_t5_act_open_req *t5req = NULL;
	struct cpl_t6_act_open_req *t6req = NULL;
	struct cpl_act_open_req6 *req6 = NULL;
	struct cpl_t5_act_open_req6 *t5req6 = NULL;
	struct cpl_t6_act_open_req6 *t6req6 = NULL;
674 675 676 677 678
	struct sk_buff *skb;
	u64 opt0;
	u32 opt2;
	unsigned int mtu_idx;
	int wscale;
679
	int win, sizev4, sizev6, wrlen;
680
	struct sockaddr_in *la = (struct sockaddr_in *)
681
				 &ep->com.local_addr;
682
	struct sockaddr_in *ra = (struct sockaddr_in *)
683
				 &ep->com.remote_addr;
684
	struct sockaddr_in6 *la6 = (struct sockaddr_in6 *)
685
				   &ep->com.local_addr;
686
	struct sockaddr_in6 *ra6 = (struct sockaddr_in6 *)
687
				   &ep->com.remote_addr;
H
Hariprasad S 已提交
688
	int ret;
689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709
	enum chip_type adapter_type = ep->com.dev->rdev.lldi.adapter_type;
	u32 isn = (prandom_u32() & ~7UL) - 1;

	switch (CHELSIO_CHIP_VERSION(adapter_type)) {
	case CHELSIO_T4:
		sizev4 = sizeof(struct cpl_act_open_req);
		sizev6 = sizeof(struct cpl_act_open_req6);
		break;
	case CHELSIO_T5:
		sizev4 = sizeof(struct cpl_t5_act_open_req);
		sizev6 = sizeof(struct cpl_t5_act_open_req6);
		break;
	case CHELSIO_T6:
		sizev4 = sizeof(struct cpl_t6_act_open_req);
		sizev6 = sizeof(struct cpl_t6_act_open_req6);
		break;
	default:
		pr_err("T%d Chip is not supported\n",
		       CHELSIO_CHIP_VERSION(adapter_type));
		return -EINVAL;
	}
710 711 712 713

	wrlen = (ep->com.remote_addr.ss_family == AF_INET) ?
			roundup(sizev4, 16) :
			roundup(sizev6, 16);
714 715 716 717 718 719 720 721 722

	PDBG("%s ep %p atid %u\n", __func__, ep, ep->atid);

	skb = get_skb(NULL, wrlen, GFP_KERNEL);
	if (!skb) {
		printk(KERN_ERR MOD "%s - failed to alloc skb.\n",
		       __func__);
		return -ENOMEM;
	}
S
Steve Wise 已提交
723
	set_wr_txq(skb, CPL_PRIORITY_SETUP, ep->ctrlq_idx);
724

725
	best_mtu(ep->com.dev->rdev.lldi.mtus, ep->mtu, &mtu_idx,
726 727
		 enable_tcp_timestamps,
		 (AF_INET == ep->com.remote_addr.ss_family) ? 0 : 1);
728
	wscale = compute_wscale(rcv_win);
729 730 731 732 733 734

	/*
	 * Specify the largest window that will fit in opt0. The
	 * remainder will be specified in the rx_data_ack.
	 */
	win = ep->rcv_win >> 10;
735 736
	if (win > RCV_BUFSIZ_M)
		win = RCV_BUFSIZ_M;
737

738
	opt0 = (nocong ? NO_CONG_F : 0) |
739
	       KEEP_ALIVE_F |
740
	       DELACK_F |
741 742 743 744 745
	       WND_SCALE_V(wscale) |
	       MSS_IDX_V(mtu_idx) |
	       L2T_IDX_V(ep->l2t->idx) |
	       TX_CHAN_V(ep->tx_chan) |
	       SMAC_SEL_V(ep->smac_idx) |
H
Hariprasad S 已提交
746
	       DSCP_V(ep->tos >> 2) |
747 748 749
	       ULP_MODE_V(ULP_MODE_TCPDDP) |
	       RCV_BUFSIZ_V(win);
	opt2 = RX_CHANNEL_V(0) |
750
	       CCTRL_ECN_V(enable_ecn) |
751
	       RSS_QUEUE_VALID_F | RSS_QUEUE_V(ep->rss_qid);
752
	if (enable_tcp_timestamps)
753
		opt2 |= TSTAMPS_EN_F;
754
	if (enable_tcp_sack)
755
		opt2 |= SACK_EN_F;
756
	if (wscale && enable_tcp_window_scaling)
757
		opt2 |= WND_SCALE_EN_F;
758 759 760 761
	if (CHELSIO_CHIP_VERSION(adapter_type) > CHELSIO_T4) {
		if (peer2peer)
			isn += 4;

762
		opt2 |= T5_OPT_2_VALID_F;
763
		opt2 |= CONG_CNTRL_V(CONG_ALG_TAHOE);
764
		opt2 |= T5_ISS_F;
765
	}
H
Hariprasad S 已提交
766 767 768 769 770

	if (ep->com.remote_addr.ss_family == AF_INET6)
		cxgb4_clip_get(ep->com.dev->rdev.lldi.ports[0],
			       (const u32 *)&la6->sin6_addr.s6_addr, 1);

771
	t4_set_arp_err_handler(skb, ep, act_open_req_arp_failure);
772

773 774 775 776
	if (ep->com.remote_addr.ss_family == AF_INET) {
		switch (CHELSIO_CHIP_VERSION(adapter_type)) {
		case CHELSIO_T4:
			req = (struct cpl_act_open_req *)skb_put(skb, wrlen);
777
			INIT_TP_WR(req, 0);
778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807
			break;
		case CHELSIO_T5:
			t5req = (struct cpl_t5_act_open_req *)skb_put(skb,
					wrlen);
			INIT_TP_WR(t5req, 0);
			req = (struct cpl_act_open_req *)t5req;
			break;
		case CHELSIO_T6:
			t6req = (struct cpl_t6_act_open_req *)skb_put(skb,
					wrlen);
			INIT_TP_WR(t6req, 0);
			req = (struct cpl_act_open_req *)t6req;
			t5req = (struct cpl_t5_act_open_req *)t6req;
			break;
		default:
			pr_err("T%d Chip is not supported\n",
			       CHELSIO_CHIP_VERSION(adapter_type));
			ret = -EINVAL;
			goto clip_release;
		}

		OPCODE_TID(req) = cpu_to_be32(MK_OPCODE_TID(CPL_ACT_OPEN_REQ,
					((ep->rss_qid<<14) | ep->atid)));
		req->local_port = la->sin_port;
		req->peer_port = ra->sin_port;
		req->local_ip = la->sin_addr.s_addr;
		req->peer_ip = ra->sin_addr.s_addr;
		req->opt0 = cpu_to_be64(opt0);

		if (is_t4(ep->com.dev->rdev.lldi.adapter_type)) {
808 809 810
			req->params = cpu_to_be32(cxgb4_select_ntuple(
						ep->com.dev->rdev.lldi.ports[0],
						ep->l2t));
811 812
			req->opt2 = cpu_to_be32(opt2);
		} else {
813 814 815 816 817 818 819 820 821 822 823
			t5req->params = cpu_to_be64(FILTER_TUPLE_V(
						cxgb4_select_ntuple(
						ep->com.dev->rdev.lldi.ports[0],
						ep->l2t)));
			t5req->rsvd = cpu_to_be32(isn);
			PDBG("%s snd_isn %u\n", __func__, t5req->rsvd);
			t5req->opt2 = cpu_to_be32(opt2);
		}
	} else {
		switch (CHELSIO_CHIP_VERSION(adapter_type)) {
		case CHELSIO_T4:
824 825
			req6 = (struct cpl_act_open_req6 *)skb_put(skb, wrlen);
			INIT_TP_WR(req6, 0);
826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857
			break;
		case CHELSIO_T5:
			t5req6 = (struct cpl_t5_act_open_req6 *)skb_put(skb,
					wrlen);
			INIT_TP_WR(t5req6, 0);
			req6 = (struct cpl_act_open_req6 *)t5req6;
			break;
		case CHELSIO_T6:
			t6req6 = (struct cpl_t6_act_open_req6 *)skb_put(skb,
					wrlen);
			INIT_TP_WR(t6req6, 0);
			req6 = (struct cpl_act_open_req6 *)t6req6;
			t5req6 = (struct cpl_t5_act_open_req6 *)t6req6;
			break;
		default:
			pr_err("T%d Chip is not supported\n",
			       CHELSIO_CHIP_VERSION(adapter_type));
			ret = -EINVAL;
			goto clip_release;
		}

		OPCODE_TID(req6) = cpu_to_be32(MK_OPCODE_TID(CPL_ACT_OPEN_REQ6,
					((ep->rss_qid<<14)|ep->atid)));
		req6->local_port = la6->sin6_port;
		req6->peer_port = ra6->sin6_port;
		req6->local_ip_hi = *((__be64 *)(la6->sin6_addr.s6_addr));
		req6->local_ip_lo = *((__be64 *)(la6->sin6_addr.s6_addr + 8));
		req6->peer_ip_hi = *((__be64 *)(ra6->sin6_addr.s6_addr));
		req6->peer_ip_lo = *((__be64 *)(ra6->sin6_addr.s6_addr + 8));
		req6->opt0 = cpu_to_be64(opt0);

		if (is_t4(ep->com.dev->rdev.lldi.adapter_type)) {
858 859 860
			req6->params = cpu_to_be32(cxgb4_select_ntuple(
						ep->com.dev->rdev.lldi.ports[0],
						ep->l2t));
861 862
			req6->opt2 = cpu_to_be32(opt2);
		} else {
863 864
			t5req6->params = cpu_to_be64(FILTER_TUPLE_V(
						cxgb4_select_ntuple(
865
						ep->com.dev->rdev.lldi.ports[0],
866
						ep->l2t)));
867 868 869
			t5req6->rsvd = cpu_to_be32(isn);
			PDBG("%s snd_isn %u\n", __func__, t5req6->rsvd);
			t5req6->opt2 = cpu_to_be32(opt2);
870
		}
871 872
	}

873
	set_bit(ACT_OPEN_REQ, &ep->com.history);
H
Hariprasad S 已提交
874
	ret = c4iw_l2t_send(&ep->com.dev->rdev, skb, ep->l2t);
875
clip_release:
H
Hariprasad S 已提交
876 877 878 879
	if (ret && ep->com.remote_addr.ss_family == AF_INET6)
		cxgb4_clip_release(ep->com.dev->rdev.lldi.ports[0],
				   (const u32 *)&la6->sin6_addr.s6_addr, 1);
	return ret;
880 881
}

882 883
static void send_mpa_req(struct c4iw_ep *ep, struct sk_buff *skb,
		u8 mpa_rev_to_use)
884 885 886 887
{
	int mpalen, wrlen;
	struct fw_ofld_tx_data_wr *req;
	struct mpa_message *mpa;
888
	struct mpa_v2_conn_params mpa_v2_params;
889 890 891 892 893 894

	PDBG("%s ep %p tid %u pd_len %d\n", __func__, ep, ep->hwtid, ep->plen);

	BUG_ON(skb_cloned(skb));

	mpalen = sizeof(*mpa) + ep->plen;
895 896
	if (mpa_rev_to_use == 2)
		mpalen += sizeof(struct mpa_v2_conn_params);
897 898 899 900 901 902 903 904 905 906 907
	wrlen = roundup(mpalen + sizeof *req, 16);
	skb = get_skb(skb, wrlen, GFP_KERNEL);
	if (!skb) {
		connect_reply_upcall(ep, -ENOMEM);
		return;
	}
	set_wr_txq(skb, CPL_PRIORITY_DATA, ep->txq_idx);

	req = (struct fw_ofld_tx_data_wr *)skb_put(skb, wrlen);
	memset(req, 0, wrlen);
	req->op_to_immdlen = cpu_to_be32(
908 909 910
		FW_WR_OP_V(FW_OFLD_TX_DATA_WR) |
		FW_WR_COMPL_F |
		FW_WR_IMMDLEN_V(mpalen));
911
	req->flowid_len16 = cpu_to_be32(
912 913
		FW_WR_FLOWID_V(ep->hwtid) |
		FW_WR_LEN16_V(wrlen >> 4));
914 915
	req->plen = cpu_to_be32(mpalen);
	req->tunnel_to_proxy = cpu_to_be32(
916 917
		FW_OFLD_TX_DATA_WR_FLUSH_F |
		FW_OFLD_TX_DATA_WR_SHOVE_F);
918 919 920 921

	mpa = (struct mpa_message *)(req + 1);
	memcpy(mpa->key, MPA_KEY_REQ, sizeof(mpa->key));
	mpa->flags = (crc_enabled ? MPA_CRC : 0) |
922 923
		     (markers_enabled ? MPA_MARKERS : 0) |
		     (mpa_rev_to_use == 2 ? MPA_ENHANCED_RDMA_CONN : 0);
924
	mpa->private_data_size = htons(ep->plen);
925
	mpa->revision = mpa_rev_to_use;
926
	if (mpa_rev_to_use == 1) {
927
		ep->tried_with_mpa_v1 = 1;
928 929
		ep->retry_with_mpa_v1 = 0;
	}
930 931

	if (mpa_rev_to_use == 2) {
932 933
		mpa->private_data_size = htons(ntohs(mpa->private_data_size) +
					       sizeof (struct mpa_v2_conn_params));
934 935
		PDBG("%s initiator ird %u ord %u\n", __func__, ep->ird,
		     ep->ord);
936 937 938 939 940 941 942 943 944 945 946 947 948 949
		mpa_v2_params.ird = htons((u16)ep->ird);
		mpa_v2_params.ord = htons((u16)ep->ord);

		if (peer2peer) {
			mpa_v2_params.ird |= htons(MPA_V2_PEER2PEER_MODEL);
			if (p2p_type == FW_RI_INIT_P2PTYPE_RDMA_WRITE)
				mpa_v2_params.ord |=
					htons(MPA_V2_RDMA_WRITE_RTR);
			else if (p2p_type == FW_RI_INIT_P2PTYPE_READ_REQ)
				mpa_v2_params.ord |=
					htons(MPA_V2_RDMA_READ_RTR);
		}
		memcpy(mpa->private_data, &mpa_v2_params,
		       sizeof(struct mpa_v2_conn_params));
950

951 952 953 954 955 956 957 958
		if (ep->plen)
			memcpy(mpa->private_data +
			       sizeof(struct mpa_v2_conn_params),
			       ep->mpa_pkt + sizeof(*mpa), ep->plen);
	} else
		if (ep->plen)
			memcpy(mpa->private_data,
					ep->mpa_pkt + sizeof(*mpa), ep->plen);
959 960 961 962 963 964 965 966 967 968 969 970

	/*
	 * Reference the mpa skb.  This ensures the data area
	 * will remain in memory until the hw acks the tx.
	 * Function fw4_ack() will deref it.
	 */
	skb_get(skb);
	t4_set_arp_err_handler(skb, NULL, arp_failure_discard);
	BUG_ON(ep->mpa_skb);
	ep->mpa_skb = skb;
	c4iw_l2t_send(&ep->com.dev->rdev, skb, ep->l2t);
	start_ep_timer(ep);
971
	__state_set(&ep->com, MPA_REQ_SENT);
972
	ep->mpa_attr.initiator = 1;
973
	ep->snd_seq += mpalen;
974 975 976 977 978 979 980 981 982
	return;
}

static int send_mpa_reject(struct c4iw_ep *ep, const void *pdata, u8 plen)
{
	int mpalen, wrlen;
	struct fw_ofld_tx_data_wr *req;
	struct mpa_message *mpa;
	struct sk_buff *skb;
983
	struct mpa_v2_conn_params mpa_v2_params;
984 985 986 987

	PDBG("%s ep %p tid %u pd_len %d\n", __func__, ep, ep->hwtid, ep->plen);

	mpalen = sizeof(*mpa) + plen;
988 989
	if (ep->mpa_attr.version == 2 && ep->mpa_attr.enhanced_rdma_conn)
		mpalen += sizeof(struct mpa_v2_conn_params);
990 991 992 993 994 995 996 997 998 999 1000 1001
	wrlen = roundup(mpalen + sizeof *req, 16);

	skb = get_skb(NULL, wrlen, GFP_KERNEL);
	if (!skb) {
		printk(KERN_ERR MOD "%s - cannot alloc skb!\n", __func__);
		return -ENOMEM;
	}
	set_wr_txq(skb, CPL_PRIORITY_DATA, ep->txq_idx);

	req = (struct fw_ofld_tx_data_wr *)skb_put(skb, wrlen);
	memset(req, 0, wrlen);
	req->op_to_immdlen = cpu_to_be32(
1002 1003 1004
		FW_WR_OP_V(FW_OFLD_TX_DATA_WR) |
		FW_WR_COMPL_F |
		FW_WR_IMMDLEN_V(mpalen));
1005
	req->flowid_len16 = cpu_to_be32(
1006 1007
		FW_WR_FLOWID_V(ep->hwtid) |
		FW_WR_LEN16_V(wrlen >> 4));
1008 1009
	req->plen = cpu_to_be32(mpalen);
	req->tunnel_to_proxy = cpu_to_be32(
1010 1011
		FW_OFLD_TX_DATA_WR_FLUSH_F |
		FW_OFLD_TX_DATA_WR_SHOVE_F);
1012 1013 1014 1015 1016

	mpa = (struct mpa_message *)(req + 1);
	memset(mpa, 0, sizeof(*mpa));
	memcpy(mpa->key, MPA_KEY_REP, sizeof(mpa->key));
	mpa->flags = MPA_REJECT;
1017
	mpa->revision = ep->mpa_attr.version;
1018
	mpa->private_data_size = htons(plen);
1019 1020 1021

	if (ep->mpa_attr.version == 2 && ep->mpa_attr.enhanced_rdma_conn) {
		mpa->flags |= MPA_ENHANCED_RDMA_CONN;
1022 1023
		mpa->private_data_size = htons(ntohs(mpa->private_data_size) +
					       sizeof (struct mpa_v2_conn_params));
1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041
		mpa_v2_params.ird = htons(((u16)ep->ird) |
					  (peer2peer ? MPA_V2_PEER2PEER_MODEL :
					   0));
		mpa_v2_params.ord = htons(((u16)ep->ord) | (peer2peer ?
					  (p2p_type ==
					   FW_RI_INIT_P2PTYPE_RDMA_WRITE ?
					   MPA_V2_RDMA_WRITE_RTR : p2p_type ==
					   FW_RI_INIT_P2PTYPE_READ_REQ ?
					   MPA_V2_RDMA_READ_RTR : 0) : 0));
		memcpy(mpa->private_data, &mpa_v2_params,
		       sizeof(struct mpa_v2_conn_params));

		if (ep->plen)
			memcpy(mpa->private_data +
			       sizeof(struct mpa_v2_conn_params), pdata, plen);
	} else
		if (plen)
			memcpy(mpa->private_data, pdata, plen);
1042 1043 1044 1045 1046 1047 1048 1049 1050 1051

	/*
	 * Reference the mpa skb again.  This ensures the data area
	 * will remain in memory until the hw acks the tx.
	 * Function fw4_ack() will deref it.
	 */
	skb_get(skb);
	set_wr_txq(skb, CPL_PRIORITY_DATA, ep->txq_idx);
	BUG_ON(ep->mpa_skb);
	ep->mpa_skb = skb;
1052
	ep->snd_seq += mpalen;
1053 1054 1055 1056 1057 1058 1059 1060 1061
	return c4iw_l2t_send(&ep->com.dev->rdev, skb, ep->l2t);
}

static int send_mpa_reply(struct c4iw_ep *ep, const void *pdata, u8 plen)
{
	int mpalen, wrlen;
	struct fw_ofld_tx_data_wr *req;
	struct mpa_message *mpa;
	struct sk_buff *skb;
1062
	struct mpa_v2_conn_params mpa_v2_params;
1063 1064 1065 1066

	PDBG("%s ep %p tid %u pd_len %d\n", __func__, ep, ep->hwtid, ep->plen);

	mpalen = sizeof(*mpa) + plen;
1067 1068
	if (ep->mpa_attr.version == 2 && ep->mpa_attr.enhanced_rdma_conn)
		mpalen += sizeof(struct mpa_v2_conn_params);
1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080
	wrlen = roundup(mpalen + sizeof *req, 16);

	skb = get_skb(NULL, wrlen, GFP_KERNEL);
	if (!skb) {
		printk(KERN_ERR MOD "%s - cannot alloc skb!\n", __func__);
		return -ENOMEM;
	}
	set_wr_txq(skb, CPL_PRIORITY_DATA, ep->txq_idx);

	req = (struct fw_ofld_tx_data_wr *) skb_put(skb, wrlen);
	memset(req, 0, wrlen);
	req->op_to_immdlen = cpu_to_be32(
1081 1082 1083
		FW_WR_OP_V(FW_OFLD_TX_DATA_WR) |
		FW_WR_COMPL_F |
		FW_WR_IMMDLEN_V(mpalen));
1084
	req->flowid_len16 = cpu_to_be32(
1085 1086
		FW_WR_FLOWID_V(ep->hwtid) |
		FW_WR_LEN16_V(wrlen >> 4));
1087 1088
	req->plen = cpu_to_be32(mpalen);
	req->tunnel_to_proxy = cpu_to_be32(
1089 1090
		FW_OFLD_TX_DATA_WR_FLUSH_F |
		FW_OFLD_TX_DATA_WR_SHOVE_F);
1091 1092 1093 1094 1095 1096

	mpa = (struct mpa_message *)(req + 1);
	memset(mpa, 0, sizeof(*mpa));
	memcpy(mpa->key, MPA_KEY_REP, sizeof(mpa->key));
	mpa->flags = (ep->mpa_attr.crc_enabled ? MPA_CRC : 0) |
		     (markers_enabled ? MPA_MARKERS : 0);
1097
	mpa->revision = ep->mpa_attr.version;
1098
	mpa->private_data_size = htons(plen);
1099 1100 1101

	if (ep->mpa_attr.version == 2 && ep->mpa_attr.enhanced_rdma_conn) {
		mpa->flags |= MPA_ENHANCED_RDMA_CONN;
1102 1103
		mpa->private_data_size = htons(ntohs(mpa->private_data_size) +
					       sizeof (struct mpa_v2_conn_params));
1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126
		mpa_v2_params.ird = htons((u16)ep->ird);
		mpa_v2_params.ord = htons((u16)ep->ord);
		if (peer2peer && (ep->mpa_attr.p2p_type !=
					FW_RI_INIT_P2PTYPE_DISABLED)) {
			mpa_v2_params.ird |= htons(MPA_V2_PEER2PEER_MODEL);

			if (p2p_type == FW_RI_INIT_P2PTYPE_RDMA_WRITE)
				mpa_v2_params.ord |=
					htons(MPA_V2_RDMA_WRITE_RTR);
			else if (p2p_type == FW_RI_INIT_P2PTYPE_READ_REQ)
				mpa_v2_params.ord |=
					htons(MPA_V2_RDMA_READ_RTR);
		}

		memcpy(mpa->private_data, &mpa_v2_params,
		       sizeof(struct mpa_v2_conn_params));

		if (ep->plen)
			memcpy(mpa->private_data +
			       sizeof(struct mpa_v2_conn_params), pdata, plen);
	} else
		if (plen)
			memcpy(mpa->private_data, pdata, plen);
1127 1128 1129 1130 1131 1132 1133 1134

	/*
	 * Reference the mpa skb.  This ensures the data area
	 * will remain in memory until the hw acks the tx.
	 * Function fw4_ack() will deref it.
	 */
	skb_get(skb);
	ep->mpa_skb = skb;
1135
	__state_set(&ep->com, MPA_REP_SENT);
1136
	ep->snd_seq += mpalen;
1137 1138 1139 1140 1141 1142 1143 1144
	return c4iw_l2t_send(&ep->com.dev->rdev, skb, ep->l2t);
}

static int act_establish(struct c4iw_dev *dev, struct sk_buff *skb)
{
	struct c4iw_ep *ep;
	struct cpl_act_establish *req = cplhdr(skb);
	unsigned int tid = GET_TID(req);
1145
	unsigned int atid = TID_TID_G(ntohl(req->tos_atid));
1146
	struct tid_info *t = dev->rdev.lldi.tids;
1147
	int ret;
1148 1149 1150 1151 1152 1153

	ep = lookup_atid(t, atid);

	PDBG("%s ep %p tid %u snd_isn %u rcv_isn %u\n", __func__, ep, tid,
	     be32_to_cpu(req->snd_isn), be32_to_cpu(req->rcv_isn));

1154
	mutex_lock(&ep->com.mutex);
1155 1156 1157 1158 1159
	dst_confirm(ep->dst);

	/* setup the hwtid for this connection */
	ep->hwtid = tid;
	cxgb4_insert_tid(t, ep, tid);
1160
	insert_handle(dev, &dev->hwtid_idr, ep, ep->hwtid);
1161 1162 1163 1164 1165 1166 1167

	ep->snd_seq = be32_to_cpu(req->snd_isn);
	ep->rcv_seq = be32_to_cpu(req->rcv_isn);

	set_emss(ep, ntohs(req->tcp_opt));

	/* dealloc the atid */
1168
	remove_handle(ep->com.dev, &ep->com.dev->atid_idr, atid);
1169
	cxgb4_free_atid(t, atid);
1170
	set_bit(ACT_ESTAB, &ep->com.history);
1171 1172

	/* start MPA negotiation */
1173 1174 1175
	ret = send_flowc(ep, NULL);
	if (ret)
		goto err;
1176 1177 1178 1179
	if (ep->retry_with_mpa_v1)
		send_mpa_req(ep, skb, 1);
	else
		send_mpa_req(ep, skb, mpa_rev);
1180
	mutex_unlock(&ep->com.mutex);
1181
	return 0;
1182 1183 1184 1185 1186
err:
	mutex_unlock(&ep->com.mutex);
	connect_reply_upcall(ep, -ENOMEM);
	c4iw_ep_disconnect(ep, 0, GFP_KERNEL);
	return 0;
1187 1188
}

1189
static void close_complete_upcall(struct c4iw_ep *ep, int status)
1190 1191 1192 1193 1194 1195
{
	struct iw_cm_event event;

	PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid);
	memset(&event, 0, sizeof(event));
	event.event = IW_CM_EVENT_CLOSE;
1196
	event.status = status;
1197 1198 1199 1200
	if (ep->com.cm_id) {
		PDBG("close complete delivered ep %p cm_id %p tid %u\n",
		     ep, ep->com.cm_id, ep->hwtid);
		ep->com.cm_id->event_handler(ep->com.cm_id, &event);
1201
		deref_cm_id(&ep->com);
1202
		set_bit(CLOSE_UPCALL, &ep->com.history);
1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216
	}
}

static void peer_close_upcall(struct c4iw_ep *ep)
{
	struct iw_cm_event event;

	PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid);
	memset(&event, 0, sizeof(event));
	event.event = IW_CM_EVENT_DISCONNECT;
	if (ep->com.cm_id) {
		PDBG("peer close delivered ep %p cm_id %p tid %u\n",
		     ep, ep->com.cm_id, ep->hwtid);
		ep->com.cm_id->event_handler(ep->com.cm_id, &event);
1217
		set_bit(DISCONN_UPCALL, &ep->com.history);
1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232
	}
}

static void peer_abort_upcall(struct c4iw_ep *ep)
{
	struct iw_cm_event event;

	PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid);
	memset(&event, 0, sizeof(event));
	event.event = IW_CM_EVENT_CLOSE;
	event.status = -ECONNRESET;
	if (ep->com.cm_id) {
		PDBG("abort delivered ep %p cm_id %p tid %u\n", ep,
		     ep->com.cm_id, ep->hwtid);
		ep->com.cm_id->event_handler(ep->com.cm_id, &event);
1233
		deref_cm_id(&ep->com);
1234
		set_bit(ABORT_UPCALL, &ep->com.history);
1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245
	}
}

static void connect_reply_upcall(struct c4iw_ep *ep, int status)
{
	struct iw_cm_event event;

	PDBG("%s ep %p tid %u status %d\n", __func__, ep, ep->hwtid, status);
	memset(&event, 0, sizeof(event));
	event.event = IW_CM_EVENT_CONNECT_REPLY;
	event.status = status;
1246 1247 1248 1249
	memcpy(&event.local_addr, &ep->com.local_addr,
	       sizeof(ep->com.local_addr));
	memcpy(&event.remote_addr, &ep->com.remote_addr,
	       sizeof(ep->com.remote_addr));
1250 1251

	if ((status == 0) || (status == -ECONNREFUSED)) {
1252 1253
		if (!ep->tried_with_mpa_v1) {
			/* this means MPA_v2 is used */
1254 1255
			event.ord = ep->ird;
			event.ird = ep->ord;
1256 1257 1258 1259 1260 1261 1262
			event.private_data_len = ep->plen -
				sizeof(struct mpa_v2_conn_params);
			event.private_data = ep->mpa_pkt +
				sizeof(struct mpa_message) +
				sizeof(struct mpa_v2_conn_params);
		} else {
			/* this means MPA_v1 is used */
1263 1264
			event.ord = cur_max_read_depth(ep->com.dev);
			event.ird = cur_max_read_depth(ep->com.dev);
1265 1266 1267 1268
			event.private_data_len = ep->plen;
			event.private_data = ep->mpa_pkt +
				sizeof(struct mpa_message);
		}
1269
	}
1270 1271 1272

	PDBG("%s ep %p tid %u status %d\n", __func__, ep,
	     ep->hwtid, status);
1273
	set_bit(CONN_RPL_UPCALL, &ep->com.history);
1274 1275
	ep->com.cm_id->event_handler(ep->com.cm_id, &event);

1276 1277
	if (status < 0)
		deref_cm_id(&ep->com);
1278 1279
}

1280
static int connect_request_upcall(struct c4iw_ep *ep)
1281 1282
{
	struct iw_cm_event event;
1283
	int ret;
1284 1285 1286 1287

	PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid);
	memset(&event, 0, sizeof(event));
	event.event = IW_CM_EVENT_CONNECT_REQUEST;
1288 1289 1290 1291
	memcpy(&event.local_addr, &ep->com.local_addr,
	       sizeof(ep->com.local_addr));
	memcpy(&event.remote_addr, &ep->com.remote_addr,
	       sizeof(ep->com.remote_addr));
1292
	event.provider_data = ep;
1293 1294 1295 1296 1297 1298 1299 1300 1301 1302
	if (!ep->tried_with_mpa_v1) {
		/* this means MPA_v2 is used */
		event.ord = ep->ord;
		event.ird = ep->ird;
		event.private_data_len = ep->plen -
			sizeof(struct mpa_v2_conn_params);
		event.private_data = ep->mpa_pkt + sizeof(struct mpa_message) +
			sizeof(struct mpa_v2_conn_params);
	} else {
		/* this means MPA_v1 is used. Send max supported */
1303 1304
		event.ord = cur_max_read_depth(ep->com.dev);
		event.ird = cur_max_read_depth(ep->com.dev);
1305 1306 1307
		event.private_data_len = ep->plen;
		event.private_data = ep->mpa_pkt + sizeof(struct mpa_message);
	}
1308 1309 1310 1311 1312
	c4iw_get_ep(&ep->com);
	ret = ep->parent_ep->com.cm_id->event_handler(ep->parent_ep->com.cm_id,
						      &event);
	if (ret)
		c4iw_put_ep(&ep->com);
1313
	set_bit(CONNREQ_UPCALL, &ep->com.history);
1314
	c4iw_put_ep(&ep->parent_ep->com);
1315
	return ret;
1316 1317 1318 1319 1320 1321 1322 1323 1324
}

static void established_upcall(struct c4iw_ep *ep)
{
	struct iw_cm_event event;

	PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid);
	memset(&event, 0, sizeof(event));
	event.event = IW_CM_EVENT_ESTABLISHED;
1325 1326
	event.ird = ep->ord;
	event.ord = ep->ird;
1327 1328 1329
	if (ep->com.cm_id) {
		PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid);
		ep->com.cm_id->event_handler(ep->com.cm_id, &event);
1330
		set_bit(ESTAB_UPCALL, &ep->com.history);
1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346
	}
}

static int update_rx_credits(struct c4iw_ep *ep, u32 credits)
{
	struct cpl_rx_data_ack *req;
	struct sk_buff *skb;
	int wrlen = roundup(sizeof *req, 16);

	PDBG("%s ep %p tid %u credits %u\n", __func__, ep, ep->hwtid, credits);
	skb = get_skb(NULL, wrlen, GFP_KERNEL);
	if (!skb) {
		printk(KERN_ERR MOD "update_rx_credits - cannot alloc skb!\n");
		return 0;
	}

1347 1348 1349 1350 1351
	/*
	 * If we couldn't specify the entire rcv window at connection setup
	 * due to the limit in the number of bits in the RCV_BUFSIZ field,
	 * then add the overage in to the credits returned.
	 */
1352 1353
	if (ep->rcv_win > RCV_BUFSIZ_M * 1024)
		credits += ep->rcv_win - RCV_BUFSIZ_M * 1024;
1354

1355 1356 1357 1358 1359
	req = (struct cpl_rx_data_ack *) skb_put(skb, wrlen);
	memset(req, 0, wrlen);
	INIT_TP_WR(req, ep->hwtid);
	OPCODE_TID(req) = cpu_to_be32(MK_OPCODE_TID(CPL_RX_DATA_ACK,
						    ep->hwtid));
1360
	req->credit_dack = cpu_to_be32(credits | RX_FORCE_ACK_F |
1361 1362
				       RX_DACK_CHANGE_F |
				       RX_DACK_MODE_V(dack_mode));
S
Steve Wise 已提交
1363
	set_wr_txq(skb, CPL_PRIORITY_ACK, ep->ctrlq_idx);
1364 1365 1366 1367
	c4iw_ofld_send(&ep->com.dev->rdev, skb);
	return credits;
}

1368 1369
#define RELAXED_IRD_NEGOTIATION 1

1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381
/*
 * process_mpa_reply - process streaming mode MPA reply
 *
 * Returns:
 *
 * 0 upon success indicating a connect request was delivered to the ULP
 * or the mpa request is incomplete but valid so far.
 *
 * 1 if a failure requires the caller to close the connection.
 *
 * 2 if a failure requires the caller to abort the connection.
 */
1382
static int process_mpa_reply(struct c4iw_ep *ep, struct sk_buff *skb)
1383 1384
{
	struct mpa_message *mpa;
1385
	struct mpa_v2_conn_params *mpa_v2_params;
1386
	u16 plen;
1387 1388
	u16 resp_ird, resp_ord;
	u8 rtr_mismatch = 0, insuff_ird = 0;
1389 1390 1391
	struct c4iw_qp_attributes attrs;
	enum c4iw_qp_attr_mask mask;
	int err;
1392
	int disconnect = 0;
1393 1394 1395 1396 1397 1398 1399 1400 1401

	PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid);

	/*
	 * If we get more than the supported amount of private data
	 * then we must fail this connection.
	 */
	if (ep->mpa_pkt_len + skb->len > sizeof(ep->mpa_pkt)) {
		err = -EINVAL;
1402
		goto err_stop_timer;
1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415
	}

	/*
	 * copy the new data into our accumulation buffer.
	 */
	skb_copy_from_linear_data(skb, &(ep->mpa_pkt[ep->mpa_pkt_len]),
				  skb->len);
	ep->mpa_pkt_len += skb->len;

	/*
	 * if we don't even have the mpa message, then bail.
	 */
	if (ep->mpa_pkt_len < sizeof(*mpa))
1416
		return 0;
1417 1418 1419
	mpa = (struct mpa_message *) ep->mpa_pkt;

	/* Validate MPA header. */
1420 1421 1422
	if (mpa->revision > mpa_rev) {
		printk(KERN_ERR MOD "%s MPA version mismatch. Local = %d,"
		       " Received = %d\n", __func__, mpa_rev, mpa->revision);
1423
		err = -EPROTO;
1424
		goto err_stop_timer;
1425 1426 1427
	}
	if (memcmp(mpa->key, MPA_KEY_REP, sizeof(mpa->key))) {
		err = -EPROTO;
1428
		goto err_stop_timer;
1429 1430 1431 1432 1433 1434 1435 1436 1437
	}

	plen = ntohs(mpa->private_data_size);

	/*
	 * Fail if there's too much private data.
	 */
	if (plen > MPA_MAX_PRIVATE_DATA) {
		err = -EPROTO;
1438
		goto err_stop_timer;
1439 1440 1441 1442 1443 1444 1445
	}

	/*
	 * If plen does not account for pkt size
	 */
	if (ep->mpa_pkt_len > (sizeof(*mpa) + plen)) {
		err = -EPROTO;
1446
		goto err_stop_timer;
1447 1448 1449 1450 1451 1452 1453 1454 1455
	}

	ep->plen = (u8) plen;

	/*
	 * If we don't have all the pdata yet, then bail.
	 * We'll continue process when more data arrives.
	 */
	if (ep->mpa_pkt_len < (sizeof(*mpa) + plen))
1456
		return 0;
1457 1458 1459

	if (mpa->flags & MPA_REJECT) {
		err = -ECONNREFUSED;
1460
		goto err_stop_timer;
1461 1462
	}

1463 1464 1465 1466 1467 1468 1469 1470
	/*
	 * Stop mpa timer.  If it expired, then
	 * we ignore the MPA reply.  process_timeout()
	 * will abort the connection.
	 */
	if (stop_ep_timer(ep))
		return 0;

1471 1472 1473 1474 1475
	/*
	 * If we get here we have accumulated the entire mpa
	 * start reply message including private data. And
	 * the MPA header is valid.
	 */
1476
	__state_set(&ep->com, FPDU_MODE);
1477 1478 1479
	ep->mpa_attr.crc_enabled = (mpa->flags & MPA_CRC) | crc_enabled ? 1 : 0;
	ep->mpa_attr.recv_marker_enabled = markers_enabled;
	ep->mpa_attr.xmit_marker_enabled = mpa->flags & MPA_MARKERS ? 1 : 0;
1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492
	ep->mpa_attr.version = mpa->revision;
	ep->mpa_attr.p2p_type = FW_RI_INIT_P2PTYPE_DISABLED;

	if (mpa->revision == 2) {
		ep->mpa_attr.enhanced_rdma_conn =
			mpa->flags & MPA_ENHANCED_RDMA_CONN ? 1 : 0;
		if (ep->mpa_attr.enhanced_rdma_conn) {
			mpa_v2_params = (struct mpa_v2_conn_params *)
				(ep->mpa_pkt + sizeof(*mpa));
			resp_ird = ntohs(mpa_v2_params->ird) &
				MPA_V2_IRD_ORD_MASK;
			resp_ord = ntohs(mpa_v2_params->ord) &
				MPA_V2_IRD_ORD_MASK;
1493 1494
			PDBG("%s responder ird %u ord %u ep ird %u ord %u\n",
			     __func__, resp_ird, resp_ord, ep->ird, ep->ord);
1495 1496 1497 1498 1499 1500

			/*
			 * This is a double-check. Ideally, below checks are
			 * not required since ird/ord stuff has been taken
			 * care of in c4iw_accept_cr
			 */
1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516
			if (ep->ird < resp_ord) {
				if (RELAXED_IRD_NEGOTIATION && resp_ord <=
				    ep->com.dev->rdev.lldi.max_ordird_qp)
					ep->ird = resp_ord;
				else
					insuff_ird = 1;
			} else if (ep->ird > resp_ord) {
				ep->ird = resp_ord;
			}
			if (ep->ord > resp_ird) {
				if (RELAXED_IRD_NEGOTIATION)
					ep->ord = resp_ird;
				else
					insuff_ird = 1;
			}
			if (insuff_ird) {
1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537
				err = -ENOMEM;
				ep->ird = resp_ord;
				ep->ord = resp_ird;
			}

			if (ntohs(mpa_v2_params->ird) &
					MPA_V2_PEER2PEER_MODEL) {
				if (ntohs(mpa_v2_params->ord) &
						MPA_V2_RDMA_WRITE_RTR)
					ep->mpa_attr.p2p_type =
						FW_RI_INIT_P2PTYPE_RDMA_WRITE;
				else if (ntohs(mpa_v2_params->ord) &
						MPA_V2_RDMA_READ_RTR)
					ep->mpa_attr.p2p_type =
						FW_RI_INIT_P2PTYPE_READ_REQ;
			}
		}
	} else if (mpa->revision == 1)
		if (peer2peer)
			ep->mpa_attr.p2p_type = p2p_type;

1538
	PDBG("%s - crc_enabled=%d, recv_marker_enabled=%d, "
1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550
	     "xmit_marker_enabled=%d, version=%d p2p_type=%d local-p2p_type = "
	     "%d\n", __func__, ep->mpa_attr.crc_enabled,
	     ep->mpa_attr.recv_marker_enabled,
	     ep->mpa_attr.xmit_marker_enabled, ep->mpa_attr.version,
	     ep->mpa_attr.p2p_type, p2p_type);

	/*
	 * If responder's RTR does not match with that of initiator, assign
	 * FW_RI_INIT_P2PTYPE_DISABLED in mpa attributes so that RTR is not
	 * generated when moving QP to RTS state.
	 * A TERM message will be sent after QP has moved to RTS state
	 */
1551
	if ((ep->mpa_attr.version == 2) && peer2peer &&
1552 1553 1554 1555
			(ep->mpa_attr.p2p_type != p2p_type)) {
		ep->mpa_attr.p2p_type = FW_RI_INIT_P2PTYPE_DISABLED;
		rtr_mismatch = 1;
	}
1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571

	attrs.mpa_attr = ep->mpa_attr;
	attrs.max_ird = ep->ird;
	attrs.max_ord = ep->ord;
	attrs.llp_stream_handle = ep;
	attrs.next_state = C4IW_QP_STATE_RTS;

	mask = C4IW_QP_ATTR_NEXT_STATE |
	    C4IW_QP_ATTR_LLP_STREAM_HANDLE | C4IW_QP_ATTR_MPA_ATTR |
	    C4IW_QP_ATTR_MAX_IRD | C4IW_QP_ATTR_MAX_ORD;

	/* bind QP and TID with INIT_WR */
	err = c4iw_modify_qp(ep->com.qp->rhp,
			     ep->com.qp, mask, &attrs, 1);
	if (err)
		goto err;
1572 1573 1574 1575 1576 1577 1578 1579 1580 1581

	/*
	 * If responder's RTR requirement did not match with what initiator
	 * supports, generate TERM message
	 */
	if (rtr_mismatch) {
		printk(KERN_ERR "%s: RTR mismatch, sending TERM\n", __func__);
		attrs.layer_etype = LAYER_MPA | DDP_LLP;
		attrs.ecode = MPA_NOMATCH_RTR;
		attrs.next_state = C4IW_QP_STATE_TERMINATE;
1582
		attrs.send_term = 1;
1583
		err = c4iw_modify_qp(ep->com.qp->rhp, ep->com.qp,
1584
				C4IW_QP_ATTR_NEXT_STATE, &attrs, 1);
1585
		err = -ENOMEM;
1586
		disconnect = 1;
1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601
		goto out;
	}

	/*
	 * Generate TERM if initiator IRD is not sufficient for responder
	 * provided ORD. Currently, we do the same behaviour even when
	 * responder provided IRD is also not sufficient as regards to
	 * initiator ORD.
	 */
	if (insuff_ird) {
		printk(KERN_ERR "%s: Insufficient IRD, sending TERM\n",
				__func__);
		attrs.layer_etype = LAYER_MPA | DDP_LLP;
		attrs.ecode = MPA_INSUFF_IRD;
		attrs.next_state = C4IW_QP_STATE_TERMINATE;
1602
		attrs.send_term = 1;
1603
		err = c4iw_modify_qp(ep->com.qp->rhp, ep->com.qp,
1604
				C4IW_QP_ATTR_NEXT_STATE, &attrs, 1);
1605
		err = -ENOMEM;
1606
		disconnect = 1;
1607 1608
		goto out;
	}
1609
	goto out;
1610 1611
err_stop_timer:
	stop_ep_timer(ep);
1612
err:
1613
	disconnect = 2;
1614 1615
out:
	connect_reply_upcall(ep, err);
1616
	return disconnect;
1617 1618
}

1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631
/*
 * process_mpa_request - process streaming mode MPA request
 *
 * Returns:
 *
 * 0 upon success indicating a connect request was delivered to the ULP
 * or the mpa request is incomplete but valid so far.
 *
 * 1 if a failure requires the caller to close the connection.
 *
 * 2 if a failure requires the caller to abort the connection.
 */
static int process_mpa_request(struct c4iw_ep *ep, struct sk_buff *skb)
1632 1633
{
	struct mpa_message *mpa;
1634
	struct mpa_v2_conn_params *mpa_v2_params;
1635 1636 1637 1638 1639 1640 1641 1642
	u16 plen;

	PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid);

	/*
	 * If we get more than the supported amount of private data
	 * then we must fail this connection.
	 */
1643 1644
	if (ep->mpa_pkt_len + skb->len > sizeof(ep->mpa_pkt))
		goto err_stop_timer;
1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659

	PDBG("%s enter (%s line %u)\n", __func__, __FILE__, __LINE__);

	/*
	 * Copy the new data into our accumulation buffer.
	 */
	skb_copy_from_linear_data(skb, &(ep->mpa_pkt[ep->mpa_pkt_len]),
				  skb->len);
	ep->mpa_pkt_len += skb->len;

	/*
	 * If we don't even have the mpa message, then bail.
	 * We'll continue process when more data arrives.
	 */
	if (ep->mpa_pkt_len < sizeof(*mpa))
1660
		return 0;
1661 1662 1663 1664 1665 1666 1667

	PDBG("%s enter (%s line %u)\n", __func__, __FILE__, __LINE__);
	mpa = (struct mpa_message *) ep->mpa_pkt;

	/*
	 * Validate MPA Header.
	 */
1668 1669 1670
	if (mpa->revision > mpa_rev) {
		printk(KERN_ERR MOD "%s MPA version mismatch. Local = %d,"
		       " Received = %d\n", __func__, mpa_rev, mpa->revision);
1671
		goto err_stop_timer;
1672 1673
	}

1674 1675
	if (memcmp(mpa->key, MPA_KEY_REQ, sizeof(mpa->key)))
		goto err_stop_timer;
1676 1677 1678 1679 1680 1681

	plen = ntohs(mpa->private_data_size);

	/*
	 * Fail if there's too much private data.
	 */
1682 1683
	if (plen > MPA_MAX_PRIVATE_DATA)
		goto err_stop_timer;
1684 1685 1686 1687

	/*
	 * If plen does not account for pkt size
	 */
1688 1689
	if (ep->mpa_pkt_len > (sizeof(*mpa) + plen))
		goto err_stop_timer;
1690 1691 1692 1693 1694 1695
	ep->plen = (u8) plen;

	/*
	 * If we don't have all the pdata yet, then bail.
	 */
	if (ep->mpa_pkt_len < (sizeof(*mpa) + plen))
1696
		return 0;
1697 1698 1699 1700 1701 1702 1703 1704 1705

	/*
	 * If we get here we have accumulated the entire mpa
	 * start reply message including private data.
	 */
	ep->mpa_attr.initiator = 0;
	ep->mpa_attr.crc_enabled = (mpa->flags & MPA_CRC) | crc_enabled ? 1 : 0;
	ep->mpa_attr.recv_marker_enabled = markers_enabled;
	ep->mpa_attr.xmit_marker_enabled = mpa->flags & MPA_MARKERS ? 1 : 0;
1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720
	ep->mpa_attr.version = mpa->revision;
	if (mpa->revision == 1)
		ep->tried_with_mpa_v1 = 1;
	ep->mpa_attr.p2p_type = FW_RI_INIT_P2PTYPE_DISABLED;

	if (mpa->revision == 2) {
		ep->mpa_attr.enhanced_rdma_conn =
			mpa->flags & MPA_ENHANCED_RDMA_CONN ? 1 : 0;
		if (ep->mpa_attr.enhanced_rdma_conn) {
			mpa_v2_params = (struct mpa_v2_conn_params *)
				(ep->mpa_pkt + sizeof(*mpa));
			ep->ird = ntohs(mpa_v2_params->ird) &
				MPA_V2_IRD_ORD_MASK;
			ep->ord = ntohs(mpa_v2_params->ord) &
				MPA_V2_IRD_ORD_MASK;
1721 1722
			PDBG("%s initiator ird %u ord %u\n", __func__, ep->ird,
			     ep->ord);
1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738
			if (ntohs(mpa_v2_params->ird) & MPA_V2_PEER2PEER_MODEL)
				if (peer2peer) {
					if (ntohs(mpa_v2_params->ord) &
							MPA_V2_RDMA_WRITE_RTR)
						ep->mpa_attr.p2p_type =
						FW_RI_INIT_P2PTYPE_RDMA_WRITE;
					else if (ntohs(mpa_v2_params->ord) &
							MPA_V2_RDMA_READ_RTR)
						ep->mpa_attr.p2p_type =
						FW_RI_INIT_P2PTYPE_READ_REQ;
				}
		}
	} else if (mpa->revision == 1)
		if (peer2peer)
			ep->mpa_attr.p2p_type = p2p_type;

1739 1740 1741 1742 1743 1744
	PDBG("%s - crc_enabled=%d, recv_marker_enabled=%d, "
	     "xmit_marker_enabled=%d, version=%d p2p_type=%d\n", __func__,
	     ep->mpa_attr.crc_enabled, ep->mpa_attr.recv_marker_enabled,
	     ep->mpa_attr.xmit_marker_enabled, ep->mpa_attr.version,
	     ep->mpa_attr.p2p_type);

1745 1746 1747 1748 1749 1750
	__state_set(&ep->com, MPA_REQ_RCVD);

	/* drive upcall */
	mutex_lock_nested(&ep->parent_ep->com.mutex, SINGLE_DEPTH_NESTING);
	if (ep->parent_ep->com.state != DEAD) {
		if (connect_request_upcall(ep))
1751
			goto err_unlock_parent;
1752 1753
	} else {
		goto err_unlock_parent;
1754
	}
1755
	mutex_unlock(&ep->parent_ep->com.mutex);
1756 1757 1758 1759 1760 1761 1762 1763 1764
	return 0;

err_unlock_parent:
	mutex_unlock(&ep->parent_ep->com.mutex);
	goto err_out;
err_stop_timer:
	(void)stop_ep_timer(ep);
err_out:
	return 2;
1765 1766 1767 1768 1769 1770 1771 1772 1773
}

static int rx_data(struct c4iw_dev *dev, struct sk_buff *skb)
{
	struct c4iw_ep *ep;
	struct cpl_rx_data *hdr = cplhdr(skb);
	unsigned int dlen = ntohs(hdr->len);
	unsigned int tid = GET_TID(hdr);
	struct tid_info *t = dev->rdev.lldi.tids;
1774
	__u8 status = hdr->status;
1775
	int disconnect = 0;
1776 1777

	ep = lookup_tid(t, tid);
1778 1779
	if (!ep)
		return 0;
1780 1781 1782
	PDBG("%s ep %p tid %u dlen %u\n", __func__, ep, ep->hwtid, dlen);
	skb_pull(skb, sizeof(*hdr));
	skb_trim(skb, dlen);
1783
	mutex_lock(&ep->com.mutex);
1784 1785 1786 1787

	/* update RX credits */
	update_rx_credits(ep, dlen);

1788
	switch (ep->com.state) {
1789
	case MPA_REQ_SENT:
1790
		ep->rcv_seq += dlen;
1791
		disconnect = process_mpa_reply(ep, skb);
1792 1793
		break;
	case MPA_REQ_WAIT:
1794
		ep->rcv_seq += dlen;
1795 1796
		process_mpa_request(ep, skb);
		break;
1797 1798 1799
	case FPDU_MODE: {
		struct c4iw_qp_attributes attrs;
		BUG_ON(!ep->com.qp);
1800
		if (status)
1801
			pr_err("%s Unexpected streaming data." \
1802 1803
			       " qpid %u ep %p state %d tid %u status %d\n",
			       __func__, ep->com.qp->wq.sq.qid, ep,
1804
			       ep->com.state, ep->hwtid, status);
1805
		attrs.next_state = C4IW_QP_STATE_TERMINATE;
1806
		c4iw_modify_qp(ep->com.qp->rhp, ep->com.qp,
1807 1808
			       C4IW_QP_ATTR_NEXT_STATE, &attrs, 1);
		disconnect = 1;
1809 1810
		break;
	}
1811 1812 1813
	default:
		break;
	}
1814
	mutex_unlock(&ep->com.mutex);
1815 1816
	if (disconnect)
		c4iw_ep_disconnect(ep, 0, GFP_KERNEL);
1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828
	return 0;
}

static int abort_rpl(struct c4iw_dev *dev, struct sk_buff *skb)
{
	struct c4iw_ep *ep;
	struct cpl_abort_rpl_rss *rpl = cplhdr(skb);
	int release = 0;
	unsigned int tid = GET_TID(rpl);
	struct tid_info *t = dev->rdev.lldi.tids;

	ep = lookup_tid(t, tid);
1829 1830 1831 1832
	if (!ep) {
		printk(KERN_WARNING MOD "Abort rpl to freed endpoint\n");
		return 0;
	}
1833
	PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid);
1834
	mutex_lock(&ep->com.mutex);
1835 1836
	switch (ep->com.state) {
	case ABORTING:
1837
		c4iw_wake_up(&ep->com.wr_wait, -ECONNRESET);
1838 1839 1840 1841 1842 1843 1844 1845
		__state_set(&ep->com, DEAD);
		release = 1;
		break;
	default:
		printk(KERN_ERR "%s ep %p state %d\n",
		     __func__, ep, ep->com.state);
		break;
	}
1846
	mutex_unlock(&ep->com.mutex);
1847 1848 1849 1850 1851 1852

	if (release)
		release_ep_resources(ep);
	return 0;
}

1853 1854 1855 1856 1857 1858
static void send_fw_act_open_req(struct c4iw_ep *ep, unsigned int atid)
{
	struct sk_buff *skb;
	struct fw_ofld_connection_wr *req;
	unsigned int mtu_idx;
	int wscale;
1859
	struct sockaddr_in *sin;
1860
	int win;
1861 1862 1863 1864

	skb = get_skb(NULL, sizeof(*req), GFP_KERNEL);
	req = (struct fw_ofld_connection_wr *)__skb_put(skb, sizeof(*req));
	memset(req, 0, sizeof(*req));
1865
	req->op_compl = htonl(WR_OP_V(FW_OFLD_CONNECTION_WR));
1866
	req->len16_pkd = htonl(FW_WR_LEN16_V(DIV_ROUND_UP(sizeof(*req), 16)));
1867 1868
	req->le.filter = cpu_to_be32(cxgb4_select_ntuple(
				     ep->com.dev->rdev.lldi.ports[0],
1869
				     ep->l2t));
1870
	sin = (struct sockaddr_in *)&ep->com.local_addr;
1871 1872
	req->le.lport = sin->sin_port;
	req->le.u.ipv4.lip = sin->sin_addr.s_addr;
1873
	sin = (struct sockaddr_in *)&ep->com.remote_addr;
1874 1875
	req->le.pport = sin->sin_port;
	req->le.u.ipv4.pip = sin->sin_addr.s_addr;
1876
	req->tcb.t_state_to_astid =
1877 1878
			htonl(FW_OFLD_CONNECTION_WR_T_STATE_V(TCP_SYN_SENT) |
			FW_OFLD_CONNECTION_WR_ASTID_V(atid));
1879
	req->tcb.cplrxdataack_cplpassacceptrpl =
1880
			htons(FW_OFLD_CONNECTION_WR_CPLRXDATAACK_F);
1881
	req->tcb.tx_max = (__force __be32) jiffies;
1882
	req->tcb.rcv_adv = htons(1);
1883
	best_mtu(ep->com.dev->rdev.lldi.mtus, ep->mtu, &mtu_idx,
1884 1885
		 enable_tcp_timestamps,
		 (AF_INET == ep->com.remote_addr.ss_family) ? 0 : 1);
1886
	wscale = compute_wscale(rcv_win);
1887 1888 1889 1890 1891 1892

	/*
	 * Specify the largest window that will fit in opt0. The
	 * remainder will be specified in the rx_data_ack.
	 */
	win = ep->rcv_win >> 10;
1893 1894
	if (win > RCV_BUFSIZ_M)
		win = RCV_BUFSIZ_M;
1895

1896 1897
	req->tcb.opt0 = (__force __be64) (TCAM_BYPASS_F |
		(nocong ? NO_CONG_F : 0) |
1898
		KEEP_ALIVE_F |
1899
		DELACK_F |
1900 1901 1902 1903 1904
		WND_SCALE_V(wscale) |
		MSS_IDX_V(mtu_idx) |
		L2T_IDX_V(ep->l2t->idx) |
		TX_CHAN_V(ep->tx_chan) |
		SMAC_SEL_V(ep->smac_idx) |
H
Hariprasad S 已提交
1905
		DSCP_V(ep->tos >> 2) |
1906 1907
		ULP_MODE_V(ULP_MODE_TCPDDP) |
		RCV_BUFSIZ_V(win));
1908 1909
	req->tcb.opt2 = (__force __be32) (PACE_V(1) |
		TX_QUEUE_V(ep->com.dev->rdev.lldi.tx_modq[ep->tx_chan]) |
1910
		RX_CHANNEL_V(0) |
1911
		CCTRL_ECN_V(enable_ecn) |
1912
		RSS_QUEUE_VALID_F | RSS_QUEUE_V(ep->rss_qid));
1913
	if (enable_tcp_timestamps)
1914
		req->tcb.opt2 |= (__force __be32)TSTAMPS_EN_F;
1915
	if (enable_tcp_sack)
1916
		req->tcb.opt2 |= (__force __be32)SACK_EN_F;
1917
	if (wscale && enable_tcp_window_scaling)
1918 1919 1920
		req->tcb.opt2 |= (__force __be32)WND_SCALE_EN_F;
	req->tcb.opt0 = cpu_to_be64((__force u64)req->tcb.opt0);
	req->tcb.opt2 = cpu_to_be32((__force u32)req->tcb.opt2);
1921 1922
	set_wr_txq(skb, CPL_PRIORITY_CONTROL, ep->ctrlq_idx);
	set_bit(ACT_OFLD_CONN, &ep->com.history);
1923 1924 1925
	c4iw_l2t_send(&ep->com.dev->rdev, skb, ep->l2t);
}

1926 1927 1928 1929 1930 1931 1932 1933 1934
/*
 * Return whether a failed active open has allocated a TID
 */
static inline int act_open_has_tid(int status)
{
	return status != CPL_ERR_TCAM_FULL && status != CPL_ERR_CONN_EXIST &&
	       status != CPL_ERR_ARP_MISS;
}

1935 1936 1937 1938 1939 1940 1941 1942 1943
/* Returns whether a CPL status conveys negative advice.
 */
static int is_neg_adv(unsigned int status)
{
	return status == CPL_ERR_RTX_NEG_ADVICE ||
	       status == CPL_ERR_PERSIST_NEG_ADVICE ||
	       status == CPL_ERR_KEEPALV_NEG_ADVICE;
}

1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957
static char *neg_adv_str(unsigned int status)
{
	switch (status) {
	case CPL_ERR_RTX_NEG_ADVICE:
		return "Retransmit timeout";
	case CPL_ERR_PERSIST_NEG_ADVICE:
		return "Persist timeout";
	case CPL_ERR_KEEPALV_NEG_ADVICE:
		return "Keepalive timeout";
	default:
		return "Unknown";
	}
}

1958 1959 1960 1961 1962 1963 1964
static void set_tcp_window(struct c4iw_ep *ep, struct port_info *pi)
{
	ep->snd_win = snd_win;
	ep->rcv_win = rcv_win;
	PDBG("%s snd_win %d rcv_win %d\n", __func__, ep->snd_win, ep->rcv_win);
}

1965 1966
#define ACT_OPEN_RETRY_COUNT 2

1967 1968
static int import_ep(struct c4iw_ep *ep, int iptype, __u8 *peer_ip,
		     struct dst_entry *dst, struct c4iw_dev *cdev,
H
Hariprasad S 已提交
1969
		     bool clear_mpa_v1, enum chip_type adapter_type, u8 tos)
1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998
{
	struct neighbour *n;
	int err, step;
	struct net_device *pdev;

	n = dst_neigh_lookup(dst, peer_ip);
	if (!n)
		return -ENODEV;

	rcu_read_lock();
	err = -ENOMEM;
	if (n->dev->flags & IFF_LOOPBACK) {
		if (iptype == 4)
			pdev = ip_dev_find(&init_net, *(__be32 *)peer_ip);
		else if (IS_ENABLED(CONFIG_IPV6))
			for_each_netdev(&init_net, pdev) {
				if (ipv6_chk_addr(&init_net,
						  (struct in6_addr *)peer_ip,
						  pdev, 1))
					break;
			}
		else
			pdev = NULL;

		if (!pdev) {
			err = -ENODEV;
			goto out;
		}
		ep->l2t = cxgb4_l2t_get(cdev->rdev.lldi.l2t,
H
Hariprasad S 已提交
1999
					n, pdev, rt_tos2priority(tos));
2000 2001 2002 2003
		if (!ep->l2t)
			goto out;
		ep->mtu = pdev->mtu;
		ep->tx_chan = cxgb4_port_chan(pdev);
2004 2005
		ep->smac_idx = cxgb4_tp_smt_idx(adapter_type,
						cxgb4_port_viid(pdev));
2006 2007 2008 2009 2010 2011 2012 2013
		step = cdev->rdev.lldi.ntxq /
			cdev->rdev.lldi.nchan;
		ep->txq_idx = cxgb4_port_idx(pdev) * step;
		step = cdev->rdev.lldi.nrxq /
			cdev->rdev.lldi.nchan;
		ep->ctrlq_idx = cxgb4_port_idx(pdev);
		ep->rss_qid = cdev->rdev.lldi.rxq_ids[
			cxgb4_port_idx(pdev) * step];
2014
		set_tcp_window(ep, (struct port_info *)netdev_priv(pdev));
2015 2016 2017 2018 2019 2020 2021 2022
		dev_put(pdev);
	} else {
		pdev = get_real_dev(n->dev);
		ep->l2t = cxgb4_l2t_get(cdev->rdev.lldi.l2t,
					n, pdev, 0);
		if (!ep->l2t)
			goto out;
		ep->mtu = dst_mtu(dst);
S
Steve Wise 已提交
2023
		ep->tx_chan = cxgb4_port_chan(pdev);
2024 2025
		ep->smac_idx = cxgb4_tp_smt_idx(adapter_type,
						cxgb4_port_viid(pdev));
2026 2027
		step = cdev->rdev.lldi.ntxq /
			cdev->rdev.lldi.nchan;
S
Steve Wise 已提交
2028 2029
		ep->txq_idx = cxgb4_port_idx(pdev) * step;
		ep->ctrlq_idx = cxgb4_port_idx(pdev);
2030 2031 2032
		step = cdev->rdev.lldi.nrxq /
			cdev->rdev.lldi.nchan;
		ep->rss_qid = cdev->rdev.lldi.rxq_ids[
S
Steve Wise 已提交
2033
			cxgb4_port_idx(pdev) * step];
2034
		set_tcp_window(ep, (struct port_info *)netdev_priv(pdev));
2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049

		if (clear_mpa_v1) {
			ep->retry_with_mpa_v1 = 0;
			ep->tried_with_mpa_v1 = 0;
		}
	}
	err = 0;
out:
	rcu_read_unlock();

	neigh_release(n);

	return err;
}

2050 2051 2052
static int c4iw_reconnect(struct c4iw_ep *ep)
{
	int err = 0;
2053
	struct sockaddr_in *laddr = (struct sockaddr_in *)
2054
				    &ep->com.cm_id->m_local_addr;
2055
	struct sockaddr_in *raddr = (struct sockaddr_in *)
2056
				    &ep->com.cm_id->m_remote_addr;
2057
	struct sockaddr_in6 *laddr6 = (struct sockaddr_in6 *)
2058
				      &ep->com.cm_id->m_local_addr;
2059
	struct sockaddr_in6 *raddr6 = (struct sockaddr_in6 *)
2060
				      &ep->com.cm_id->m_remote_addr;
2061 2062
	int iptype;
	__u8 *ra;
2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078

	PDBG("%s qp %p cm_id %p\n", __func__, ep->com.qp, ep->com.cm_id);
	init_timer(&ep->timer);

	/*
	 * Allocate an active TID to initiate a TCP connection.
	 */
	ep->atid = cxgb4_alloc_atid(ep->com.dev->rdev.lldi.tids, ep);
	if (ep->atid == -1) {
		pr_err("%s - cannot alloc atid.\n", __func__);
		err = -ENOMEM;
		goto fail2;
	}
	insert_handle(ep->com.dev, &ep->com.dev->atid_idr, ep, ep->atid);

	/* find a route */
2079
	if (ep->com.cm_id->m_local_addr.ss_family == AF_INET) {
2080 2081
		ep->dst = find_route(ep->com.dev, laddr->sin_addr.s_addr,
				     raddr->sin_addr.s_addr, laddr->sin_port,
H
Hariprasad S 已提交
2082
				     raddr->sin_port, ep->com.cm_id->tos);
2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093
		iptype = 4;
		ra = (__u8 *)&raddr->sin_addr;
	} else {
		ep->dst = find_route6(ep->com.dev, laddr6->sin6_addr.s6_addr,
				      raddr6->sin6_addr.s6_addr,
				      laddr6->sin6_port, raddr6->sin6_port, 0,
				      raddr6->sin6_scope_id);
		iptype = 6;
		ra = (__u8 *)&raddr6->sin6_addr;
	}
	if (!ep->dst) {
2094 2095 2096 2097
		pr_err("%s - cannot find route.\n", __func__);
		err = -EHOSTUNREACH;
		goto fail3;
	}
2098
	err = import_ep(ep, iptype, ra, ep->dst, ep->com.dev, false,
H
Hariprasad S 已提交
2099 2100
			ep->com.dev->rdev.lldi.adapter_type,
			ep->com.cm_id->tos);
2101
	if (err) {
2102 2103 2104 2105 2106 2107 2108 2109 2110
		pr_err("%s - cannot alloc l2e.\n", __func__);
		goto fail4;
	}

	PDBG("%s txq_idx %u tx_chan %u smac_idx %u rss_qid %u l2t_idx %u\n",
	     __func__, ep->txq_idx, ep->tx_chan, ep->smac_idx, ep->rss_qid,
	     ep->l2t->idx);

	state_set(&ep->com, CONNECTING);
H
Hariprasad S 已提交
2111
	ep->tos = ep->com.cm_id->tos;
2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136

	/* send connect request to rnic */
	err = send_connect(ep);
	if (!err)
		goto out;

	cxgb4_l2t_release(ep->l2t);
fail4:
	dst_release(ep->dst);
fail3:
	remove_handle(ep->com.dev, &ep->com.dev->atid_idr, ep->atid);
	cxgb4_free_atid(ep->com.dev->rdev.lldi.tids, ep->atid);
fail2:
	/*
	 * remember to send notification to upper layer.
	 * We are in here so the upper layer is not aware that this is
	 * re-connect attempt and so, upper layer is still waiting for
	 * response of 1st connect request.
	 */
	connect_reply_upcall(ep, -ECONNRESET);
	c4iw_put_ep(&ep->com);
out:
	return err;
}

2137 2138 2139 2140
static int act_open_rpl(struct c4iw_dev *dev, struct sk_buff *skb)
{
	struct c4iw_ep *ep;
	struct cpl_act_open_rpl *rpl = cplhdr(skb);
2141 2142
	unsigned int atid = TID_TID_G(AOPEN_ATID_G(
				      ntohl(rpl->atid_status)));
2143
	struct tid_info *t = dev->rdev.lldi.tids;
2144
	int status = AOPEN_STATUS_G(ntohl(rpl->atid_status));
2145 2146 2147 2148
	struct sockaddr_in *la;
	struct sockaddr_in *ra;
	struct sockaddr_in6 *la6;
	struct sockaddr_in6 *ra6;
2149 2150

	ep = lookup_atid(t, atid);
2151 2152 2153 2154
	la = (struct sockaddr_in *)&ep->com.local_addr;
	ra = (struct sockaddr_in *)&ep->com.remote_addr;
	la6 = (struct sockaddr_in6 *)&ep->com.local_addr;
	ra6 = (struct sockaddr_in6 *)&ep->com.remote_addr;
2155 2156 2157 2158

	PDBG("%s ep %p atid %u status %u errno %d\n", __func__, ep, atid,
	     status, status2errno(status));

2159
	if (is_neg_adv(status)) {
2160 2161 2162 2163 2164 2165
		PDBG("%s Connection problems for atid %u status %u (%s)\n",
		     __func__, atid, status, neg_adv_str(status));
		ep->stats.connect_neg_adv++;
		mutex_lock(&dev->rdev.stats.lock);
		dev->rdev.stats.neg_adv++;
		mutex_unlock(&dev->rdev.stats.lock);
2166 2167 2168
		return 0;
	}

2169 2170
	set_bit(ACT_OPEN_RPL, &ep->com.history);

2171 2172 2173 2174 2175 2176 2177
	/*
	 * Log interesting failures.
	 */
	switch (status) {
	case CPL_ERR_CONN_RESET:
	case CPL_ERR_CONN_TIMEDOUT:
		break;
2178
	case CPL_ERR_TCAM_FULL:
2179
		mutex_lock(&dev->rdev.stats.lock);
2180
		dev->rdev.stats.tcam_full++;
2181 2182 2183
		mutex_unlock(&dev->rdev.stats.lock);
		if (ep->com.local_addr.ss_family == AF_INET &&
		    dev->rdev.lldi.enable_fw_ofld_conn) {
2184
			send_fw_act_open_req(ep,
2185
					     TID_TID_G(AOPEN_ATID_G(
2186 2187 2188 2189 2190 2191 2192
					     ntohl(rpl->atid_status))));
			return 0;
		}
		break;
	case CPL_ERR_CONN_EXIST:
		if (ep->retry_count++ < ACT_OPEN_RETRY_COUNT) {
			set_bit(ACT_RETRY_INUSE, &ep->com.history);
H
Hariprasad S 已提交
2193 2194 2195
			if (ep->com.remote_addr.ss_family == AF_INET6) {
				struct sockaddr_in6 *sin6 =
						(struct sockaddr_in6 *)
2196
						&ep->com.local_addr;
H
Hariprasad S 已提交
2197 2198 2199 2200 2201
				cxgb4_clip_release(
						ep->com.dev->rdev.lldi.ports[0],
						(const u32 *)
						&sin6->sin6_addr.s6_addr, 1);
			}
2202 2203 2204 2205 2206 2207 2208 2209
			remove_handle(ep->com.dev, &ep->com.dev->atid_idr,
					atid);
			cxgb4_free_atid(t, atid);
			dst_release(ep->dst);
			cxgb4_l2t_release(ep->l2t);
			c4iw_reconnect(ep);
			return 0;
		}
2210
		break;
2211
	default:
2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222
		if (ep->com.local_addr.ss_family == AF_INET) {
			pr_info("Active open failure - atid %u status %u errno %d %pI4:%u->%pI4:%u\n",
				atid, status, status2errno(status),
				&la->sin_addr.s_addr, ntohs(la->sin_port),
				&ra->sin_addr.s_addr, ntohs(ra->sin_port));
		} else {
			pr_info("Active open failure - atid %u status %u errno %d %pI6:%u->%pI6:%u\n",
				atid, status, status2errno(status),
				la6->sin6_addr.s6_addr, ntohs(la6->sin6_port),
				ra6->sin6_addr.s6_addr, ntohs(ra6->sin6_port));
		}
2223 2224 2225
		break;
	}

2226 2227 2228
	connect_reply_upcall(ep, status2errno(status));
	state_set(&ep->com, DEAD);

H
Hariprasad S 已提交
2229 2230
	if (ep->com.remote_addr.ss_family == AF_INET6) {
		struct sockaddr_in6 *sin6 =
2231
			(struct sockaddr_in6 *)&ep->com.local_addr;
H
Hariprasad S 已提交
2232 2233 2234
		cxgb4_clip_release(ep->com.dev->rdev.lldi.ports[0],
				   (const u32 *)&sin6->sin6_addr.s6_addr, 1);
	}
2235 2236 2237
	if (status && act_open_has_tid(status))
		cxgb4_remove_tid(ep->com.dev->rdev.lldi.tids, 0, GET_TID(rpl));

2238
	remove_handle(ep->com.dev, &ep->com.dev->atid_idr, atid);
2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254
	cxgb4_free_atid(t, atid);
	dst_release(ep->dst);
	cxgb4_l2t_release(ep->l2t);
	c4iw_put_ep(&ep->com);

	return 0;
}

static int pass_open_rpl(struct c4iw_dev *dev, struct sk_buff *skb)
{
	struct cpl_pass_open_rpl *rpl = cplhdr(skb);
	struct tid_info *t = dev->rdev.lldi.tids;
	unsigned int stid = GET_TID(rpl);
	struct c4iw_listen_ep *ep = lookup_stid(t, stid);

	if (!ep) {
2255 2256
		PDBG("%s stid %d lookup failure!\n", __func__, stid);
		goto out;
2257 2258 2259
	}
	PDBG("%s ep %p status %d error %d\n", __func__, ep,
	     rpl->status, status2errno(rpl->status));
2260
	c4iw_wake_up(&ep->com.wr_wait, status2errno(rpl->status));
2261

2262
out:
2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273
	return 0;
}

static int close_listsrv_rpl(struct c4iw_dev *dev, struct sk_buff *skb)
{
	struct cpl_close_listsvr_rpl *rpl = cplhdr(skb);
	struct tid_info *t = dev->rdev.lldi.tids;
	unsigned int stid = GET_TID(rpl);
	struct c4iw_listen_ep *ep = lookup_stid(t, stid);

	PDBG("%s ep %p\n", __func__, ep);
2274
	c4iw_wake_up(&ep->com.wr_wait, status2errno(rpl->status));
2275 2276 2277
	return 0;
}

2278 2279
static int accept_cr(struct c4iw_ep *ep, struct sk_buff *skb,
		     struct cpl_pass_accept_req *req)
2280 2281 2282 2283 2284 2285
{
	struct cpl_pass_accept_rpl *rpl;
	unsigned int mtu_idx;
	u64 opt0;
	u32 opt2;
	int wscale;
2286
	struct cpl_t5_pass_accept_rpl *rpl5 = NULL;
2287
	int win;
2288
	enum chip_type adapter_type = ep->com.dev->rdev.lldi.adapter_type;
2289 2290 2291

	PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid);
	BUG_ON(skb_cloned(skb));
2292

2293
	skb_get(skb);
2294
	rpl = cplhdr(skb);
2295
	if (!is_t4(adapter_type)) {
2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306
		skb_trim(skb, roundup(sizeof(*rpl5), 16));
		rpl5 = (void *)rpl;
		INIT_TP_WR(rpl5, ep->hwtid);
	} else {
		skb_trim(skb, sizeof(*rpl));
		INIT_TP_WR(rpl, ep->hwtid);
	}
	OPCODE_TID(rpl) = cpu_to_be32(MK_OPCODE_TID(CPL_PASS_ACCEPT_RPL,
						    ep->hwtid));

	best_mtu(ep->com.dev->rdev.lldi.mtus, ep->mtu, &mtu_idx,
2307 2308
		 enable_tcp_timestamps && req->tcpopt.tstamp,
		 (AF_INET == ep->com.remote_addr.ss_family) ? 0 : 1);
2309
	wscale = compute_wscale(rcv_win);
2310 2311 2312 2313 2314 2315

	/*
	 * Specify the largest window that will fit in opt0. The
	 * remainder will be specified in the rx_data_ack.
	 */
	win = ep->rcv_win >> 10;
2316 2317
	if (win > RCV_BUFSIZ_M)
		win = RCV_BUFSIZ_M;
2318
	opt0 = (nocong ? NO_CONG_F : 0) |
2319
	       KEEP_ALIVE_F |
2320
	       DELACK_F |
2321 2322 2323 2324 2325
	       WND_SCALE_V(wscale) |
	       MSS_IDX_V(mtu_idx) |
	       L2T_IDX_V(ep->l2t->idx) |
	       TX_CHAN_V(ep->tx_chan) |
	       SMAC_SEL_V(ep->smac_idx) |
2326
	       DSCP_V(ep->tos >> 2) |
2327 2328 2329 2330
	       ULP_MODE_V(ULP_MODE_TCPDDP) |
	       RCV_BUFSIZ_V(win);
	opt2 = RX_CHANNEL_V(0) |
	       RSS_QUEUE_VALID_F | RSS_QUEUE_V(ep->rss_qid);
2331 2332

	if (enable_tcp_timestamps && req->tcpopt.tstamp)
2333
		opt2 |= TSTAMPS_EN_F;
2334
	if (enable_tcp_sack && req->tcpopt.sack)
2335
		opt2 |= SACK_EN_F;
2336
	if (wscale && enable_tcp_window_scaling)
2337
		opt2 |= WND_SCALE_EN_F;
2338 2339 2340 2341
	if (enable_ecn) {
		const struct tcphdr *tcph;
		u32 hlen = ntohl(req->hdr_len);

2342 2343 2344 2345 2346 2347
		if (CHELSIO_CHIP_VERSION(adapter_type) <= CHELSIO_T5)
			tcph = (const void *)(req + 1) + ETH_HDR_LEN_G(hlen) +
				IP_HDR_LEN_G(hlen);
		else
			tcph = (const void *)(req + 1) +
				T6_ETH_HDR_LEN_G(hlen) + T6_IP_HDR_LEN_G(hlen);
2348
		if (tcph->ece && tcph->cwr)
2349
			opt2 |= CCTRL_ECN_V(1);
2350
	}
2351
	if (CHELSIO_CHIP_VERSION(adapter_type) > CHELSIO_T4) {
2352
		u32 isn = (prandom_u32() & ~7UL) - 1;
2353
		opt2 |= T5_OPT_2_VALID_F;
2354
		opt2 |= CONG_CNTRL_V(CONG_ALG_TAHOE);
2355
		opt2 |= T5_ISS_F;
2356 2357 2358 2359 2360 2361
		rpl5 = (void *)rpl;
		memset(&rpl5->iss, 0, roundup(sizeof(*rpl5)-sizeof(*rpl), 16));
		if (peer2peer)
			isn += 4;
		rpl5->iss = cpu_to_be32(isn);
		PDBG("%s iss %u\n", __func__, be32_to_cpu(rpl5->iss));
2362
	}
2363 2364 2365

	rpl->opt0 = cpu_to_be64(opt0);
	rpl->opt2 = cpu_to_be32(opt2);
S
Steve Wise 已提交
2366
	set_wr_txq(skb, CPL_PRIORITY_SETUP, ep->ctrlq_idx);
2367
	t4_set_arp_err_handler(skb, ep, pass_accept_rpl_arp_failure);
2368

2369
	return c4iw_l2t_send(&ep->com.dev->rdev, skb, ep->l2t);
2370 2371
}

2372
static void reject_cr(struct c4iw_dev *dev, u32 hwtid, struct sk_buff *skb)
2373
{
2374
	PDBG("%s c4iw_dev %p tid %u\n", __func__, dev, hwtid);
2375 2376 2377 2378 2379 2380
	BUG_ON(skb_cloned(skb));
	skb_trim(skb, sizeof(struct cpl_tid_release));
	release_tid(&dev->rdev, hwtid, skb);
	return;
}

2381 2382
static void get_4tuple(struct cpl_pass_accept_req *req, enum chip_type type,
		       int *iptype, __u8 *local_ip, __u8 *peer_ip,
2383 2384
		       __be16 *local_port, __be16 *peer_port)
{
2385 2386 2387 2388 2389 2390
	int eth_len = (CHELSIO_CHIP_VERSION(type) <= CHELSIO_T5) ?
		      ETH_HDR_LEN_G(be32_to_cpu(req->hdr_len)) :
		      T6_ETH_HDR_LEN_G(be32_to_cpu(req->hdr_len));
	int ip_len = (CHELSIO_CHIP_VERSION(type) <= CHELSIO_T5) ?
		     IP_HDR_LEN_G(be32_to_cpu(req->hdr_len)) :
		     T6_IP_HDR_LEN_G(be32_to_cpu(req->hdr_len));
2391
	struct iphdr *ip = (struct iphdr *)((u8 *)(req + 1) + eth_len);
2392
	struct ipv6hdr *ip6 = (struct ipv6hdr *)((u8 *)(req + 1) + eth_len);
2393 2394 2395
	struct tcphdr *tcp = (struct tcphdr *)
			     ((u8 *)(req + 1) + eth_len + ip_len);

2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410
	if (ip->version == 4) {
		PDBG("%s saddr 0x%x daddr 0x%x sport %u dport %u\n", __func__,
		     ntohl(ip->saddr), ntohl(ip->daddr), ntohs(tcp->source),
		     ntohs(tcp->dest));
		*iptype = 4;
		memcpy(peer_ip, &ip->saddr, 4);
		memcpy(local_ip, &ip->daddr, 4);
	} else {
		PDBG("%s saddr %pI6 daddr %pI6 sport %u dport %u\n", __func__,
		     ip6->saddr.s6_addr, ip6->daddr.s6_addr, ntohs(tcp->source),
		     ntohs(tcp->dest));
		*iptype = 6;
		memcpy(peer_ip, ip6->saddr.s6_addr, 16);
		memcpy(local_ip, ip6->daddr.s6_addr, 16);
	}
2411 2412 2413 2414 2415 2416 2417 2418
	*peer_port = tcp->source;
	*local_port = tcp->dest;

	return;
}

static int pass_accept_req(struct c4iw_dev *dev, struct sk_buff *skb)
{
2419
	struct c4iw_ep *child_ep = NULL, *parent_ep;
2420
	struct cpl_pass_accept_req *req = cplhdr(skb);
2421
	unsigned int stid = PASS_OPEN_TID_G(ntohl(req->tos_stid));
2422 2423 2424
	struct tid_info *t = dev->rdev.lldi.tids;
	unsigned int hwtid = GET_TID(req);
	struct dst_entry *dst;
2425
	__u8 local_ip[16], peer_ip[16];
2426
	__be16 local_port, peer_port;
H
Hariprasad S 已提交
2427
	struct sockaddr_in6 *sin6;
2428
	int err;
2429
	u16 peer_mss = ntohs(req->tcpopt.mss);
2430
	int iptype;
2431
	unsigned short hdrs;
H
Hariprasad S 已提交
2432
	u8 tos = PASS_OPEN_TOS_G(ntohl(req->tos_stid));
2433 2434

	parent_ep = lookup_stid(t, stid);
2435 2436 2437 2438 2439
	if (!parent_ep) {
		PDBG("%s connect request on invalid stid %d\n", __func__, stid);
		goto reject;
	}

2440
	if (state_read(&parent_ep->com) != LISTEN) {
2441
		PDBG("%s - listening ep not in LISTEN\n", __func__);
2442 2443 2444
		goto reject;
	}

2445 2446
	get_4tuple(req, parent_ep->com.dev->rdev.lldi.adapter_type, &iptype,
		   local_ip, peer_ip, &local_port, &peer_port);
2447

2448
	/* Find output route */
2449 2450 2451 2452 2453 2454 2455
	if (iptype == 4)  {
		PDBG("%s parent ep %p hwtid %u laddr %pI4 raddr %pI4 lport %d rport %d peer_mss %d\n"
		     , __func__, parent_ep, hwtid,
		     local_ip, peer_ip, ntohs(local_port),
		     ntohs(peer_port), peer_mss);
		dst = find_route(dev, *(__be32 *)local_ip, *(__be32 *)peer_ip,
				 local_port, peer_port,
H
Hariprasad S 已提交
2456
				 tos);
2457 2458 2459 2460 2461 2462
	} else {
		PDBG("%s parent ep %p hwtid %u laddr %pI6 raddr %pI6 lport %d rport %d peer_mss %d\n"
		     , __func__, parent_ep, hwtid,
		     local_ip, peer_ip, ntohs(local_port),
		     ntohs(peer_port), peer_mss);
		dst = find_route6(dev, local_ip, peer_ip, local_port, peer_port,
2463
				  PASS_OPEN_TOS_G(ntohl(req->tos_stid)),
2464 2465 2466 2467
				  ((struct sockaddr_in6 *)
				  &parent_ep->com.local_addr)->sin6_scope_id);
	}
	if (!dst) {
2468 2469 2470 2471
		printk(KERN_ERR MOD "%s - failed to find dst entry!\n",
		       __func__);
		goto reject;
	}
2472 2473 2474 2475

	child_ep = alloc_ep(sizeof(*child_ep), GFP_KERNEL);
	if (!child_ep) {
		printk(KERN_ERR MOD "%s - failed to allocate ep entry!\n",
2476 2477 2478 2479 2480
		       __func__);
		dst_release(dst);
		goto reject;
	}

2481
	err = import_ep(child_ep, iptype, peer_ip, dst, dev, false,
H
Hariprasad S 已提交
2482
			parent_ep->com.dev->rdev.lldi.adapter_type, tos);
2483 2484
	if (err) {
		printk(KERN_ERR MOD "%s - failed to allocate l2t entry!\n",
2485 2486
		       __func__);
		dst_release(dst);
2487
		kfree(child_ep);
2488 2489
		goto reject;
	}
2490

2491 2492 2493 2494
	hdrs = sizeof(struct iphdr) + sizeof(struct tcphdr) +
	       ((enable_tcp_timestamps && req->tcpopt.tstamp) ? 12 : 0);
	if (peer_mss && child_ep->mtu > (peer_mss + hdrs))
		child_ep->mtu = peer_mss + hdrs;
2495

2496 2497 2498
	state_set(&child_ep->com, CONNECTING);
	child_ep->com.dev = dev;
	child_ep->com.cm_id = NULL;
2499

2500 2501
	if (iptype == 4) {
		struct sockaddr_in *sin = (struct sockaddr_in *)
2502
			&child_ep->com.local_addr;
2503

2504 2505 2506
		sin->sin_family = PF_INET;
		sin->sin_port = local_port;
		sin->sin_addr.s_addr = *(__be32 *)local_ip;
2507 2508 2509 2510 2511 2512 2513

		sin = (struct sockaddr_in *)&child_ep->com.local_addr;
		sin->sin_family = PF_INET;
		sin->sin_port = ((struct sockaddr_in *)
				 &parent_ep->com.local_addr)->sin_port;
		sin->sin_addr.s_addr = *(__be32 *)local_ip;

2514
		sin = (struct sockaddr_in *)&child_ep->com.remote_addr;
2515 2516 2517 2518
		sin->sin_family = PF_INET;
		sin->sin_port = peer_port;
		sin->sin_addr.s_addr = *(__be32 *)peer_ip;
	} else {
2519
		sin6 = (struct sockaddr_in6 *)&child_ep->com.local_addr;
2520 2521 2522
		sin6->sin6_family = PF_INET6;
		sin6->sin6_port = local_port;
		memcpy(sin6->sin6_addr.s6_addr, local_ip, 16);
2523 2524 2525 2526 2527 2528 2529

		sin6 = (struct sockaddr_in6 *)&child_ep->com.local_addr;
		sin6->sin6_family = PF_INET6;
		sin6->sin6_port = ((struct sockaddr_in6 *)
				   &parent_ep->com.local_addr)->sin6_port;
		memcpy(sin6->sin6_addr.s6_addr, local_ip, 16);

2530
		sin6 = (struct sockaddr_in6 *)&child_ep->com.remote_addr;
2531 2532 2533 2534
		sin6->sin6_family = PF_INET6;
		sin6->sin6_port = peer_port;
		memcpy(sin6->sin6_addr.s6_addr, peer_ip, 16);
	}
2535

2536 2537
	c4iw_get_ep(&parent_ep->com);
	child_ep->parent_ep = parent_ep;
H
Hariprasad S 已提交
2538
	child_ep->tos = tos;
2539 2540 2541 2542
	child_ep->dst = dst;
	child_ep->hwtid = hwtid;

	PDBG("%s tx_chan %u smac_idx %u rss_qid %u\n", __func__,
2543
	     child_ep->tx_chan, child_ep->smac_idx, child_ep->rss_qid);
2544 2545 2546

	init_timer(&child_ep->timer);
	cxgb4_insert_tid(t, child_ep, hwtid);
2547
	insert_handle(dev, &dev->hwtid_idr, child_ep, child_ep->hwtid);
2548 2549 2550 2551 2552 2553
	if (accept_cr(child_ep, skb, req)) {
		c4iw_put_ep(&parent_ep->com);
		release_ep_resources(child_ep);
	} else {
		set_bit(PASS_ACCEPT_REQ, &child_ep->com.history);
	}
H
Hariprasad S 已提交
2554
	if (iptype == 6) {
2555
		sin6 = (struct sockaddr_in6 *)&child_ep->com.local_addr;
H
Hariprasad S 已提交
2556 2557 2558
		cxgb4_clip_get(child_ep->com.dev->rdev.lldi.ports[0],
			       (const u32 *)&sin6->sin6_addr.s6_addr, 1);
	}
2559 2560
	goto out;
reject:
2561
	reject_cr(dev, hwtid, skb);
2562 2563 2564 2565 2566 2567 2568 2569 2570 2571
out:
	return 0;
}

static int pass_establish(struct c4iw_dev *dev, struct sk_buff *skb)
{
	struct c4iw_ep *ep;
	struct cpl_pass_establish *req = cplhdr(skb);
	struct tid_info *t = dev->rdev.lldi.tids;
	unsigned int tid = GET_TID(req);
2572
	int ret;
2573 2574 2575 2576 2577 2578

	ep = lookup_tid(t, tid);
	PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid);
	ep->snd_seq = be32_to_cpu(req->snd_isn);
	ep->rcv_seq = be32_to_cpu(req->rcv_isn);

2579 2580 2581
	PDBG("%s ep %p hwtid %u tcp_opt 0x%02x\n", __func__, ep, tid,
	     ntohs(req->tcp_opt));

2582 2583 2584
	set_emss(ep, ntohs(req->tcp_opt));

	dst_confirm(ep->dst);
2585 2586
	mutex_lock(&ep->com.mutex);
	ep->com.state = MPA_REQ_WAIT;
2587
	start_ep_timer(ep);
2588
	set_bit(PASS_ESTAB, &ep->com.history);
2589 2590 2591 2592
	ret = send_flowc(ep, skb);
	mutex_unlock(&ep->com.mutex);
	if (ret)
		c4iw_ep_disconnect(ep, 1, GFP_KERNEL);
2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605

	return 0;
}

static int peer_close(struct c4iw_dev *dev, struct sk_buff *skb)
{
	struct cpl_peer_close *hdr = cplhdr(skb);
	struct c4iw_ep *ep;
	struct c4iw_qp_attributes attrs;
	int disconnect = 1;
	int release = 0;
	struct tid_info *t = dev->rdev.lldi.tids;
	unsigned int tid = GET_TID(hdr);
S
Steve Wise 已提交
2606
	int ret;
2607 2608 2609 2610 2611

	ep = lookup_tid(t, tid);
	PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid);
	dst_confirm(ep->dst);

2612
	set_bit(PEER_CLOSE, &ep->com.history);
2613
	mutex_lock(&ep->com.mutex);
2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631
	switch (ep->com.state) {
	case MPA_REQ_WAIT:
		__state_set(&ep->com, CLOSING);
		break;
	case MPA_REQ_SENT:
		__state_set(&ep->com, CLOSING);
		connect_reply_upcall(ep, -ECONNRESET);
		break;
	case MPA_REQ_RCVD:

		/*
		 * We're gonna mark this puppy DEAD, but keep
		 * the reference on it until the ULP accepts or
		 * rejects the CR. Also wake up anyone waiting
		 * in rdma connection migration (see c4iw_accept_cr()).
		 */
		__state_set(&ep->com, CLOSING);
		PDBG("waking up ep %p tid %u\n", ep, ep->hwtid);
2632
		c4iw_wake_up(&ep->com.wr_wait, -ECONNRESET);
2633 2634 2635 2636
		break;
	case MPA_REP_SENT:
		__state_set(&ep->com, CLOSING);
		PDBG("waking up ep %p tid %u\n", ep, ep->hwtid);
2637
		c4iw_wake_up(&ep->com.wr_wait, -ECONNRESET);
2638 2639
		break;
	case FPDU_MODE:
2640
		start_ep_timer(ep);
2641
		__state_set(&ep->com, CLOSING);
2642
		attrs.next_state = C4IW_QP_STATE_CLOSING;
S
Steve Wise 已提交
2643
		ret = c4iw_modify_qp(ep->com.qp->rhp, ep->com.qp,
2644
				       C4IW_QP_ATTR_NEXT_STATE, &attrs, 1);
S
Steve Wise 已提交
2645 2646 2647 2648
		if (ret != -ECONNRESET) {
			peer_close_upcall(ep);
			disconnect = 1;
		}
2649 2650 2651 2652 2653 2654 2655 2656 2657
		break;
	case ABORTING:
		disconnect = 0;
		break;
	case CLOSING:
		__state_set(&ep->com, MORIBUND);
		disconnect = 0;
		break;
	case MORIBUND:
S
Steve Wise 已提交
2658
		(void)stop_ep_timer(ep);
2659 2660 2661 2662 2663
		if (ep->com.cm_id && ep->com.qp) {
			attrs.next_state = C4IW_QP_STATE_IDLE;
			c4iw_modify_qp(ep->com.qp->rhp, ep->com.qp,
				       C4IW_QP_ATTR_NEXT_STATE, &attrs, 1);
		}
2664
		close_complete_upcall(ep, 0);
2665 2666 2667 2668 2669 2670 2671 2672 2673 2674
		__state_set(&ep->com, DEAD);
		release = 1;
		disconnect = 0;
		break;
	case DEAD:
		disconnect = 0;
		break;
	default:
		BUG_ON(1);
	}
2675
	mutex_unlock(&ep->com.mutex);
2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695
	if (disconnect)
		c4iw_ep_disconnect(ep, 0, GFP_KERNEL);
	if (release)
		release_ep_resources(ep);
	return 0;
}

static int peer_abort(struct c4iw_dev *dev, struct sk_buff *skb)
{
	struct cpl_abort_req_rss *req = cplhdr(skb);
	struct c4iw_ep *ep;
	struct cpl_abort_rpl *rpl;
	struct sk_buff *rpl_skb;
	struct c4iw_qp_attributes attrs;
	int ret;
	int release = 0;
	struct tid_info *t = dev->rdev.lldi.tids;
	unsigned int tid = GET_TID(req);

	ep = lookup_tid(t, tid);
2696
	if (is_neg_adv(req->status)) {
2697 2698 2699 2700 2701 2702 2703
		PDBG("%s Negative advice on abort- tid %u status %d (%s)\n",
		     __func__, ep->hwtid, req->status,
		     neg_adv_str(req->status));
		ep->stats.abort_neg_adv++;
		mutex_lock(&dev->rdev.stats.lock);
		dev->rdev.stats.neg_adv++;
		mutex_unlock(&dev->rdev.stats.lock);
2704 2705 2706 2707
		return 0;
	}
	PDBG("%s ep %p tid %u state %u\n", __func__, ep, ep->hwtid,
	     ep->com.state);
2708
	set_bit(PEER_ABORT, &ep->com.history);
2709 2710 2711

	/*
	 * Wake up any threads in rdma_init() or rdma_fini().
2712 2713
	 * However, this is not needed if com state is just
	 * MPA_REQ_SENT
2714
	 */
2715 2716
	if (ep->com.state != MPA_REQ_SENT)
		c4iw_wake_up(&ep->com.wr_wait, -ECONNRESET);
2717 2718

	mutex_lock(&ep->com.mutex);
2719 2720
	switch (ep->com.state) {
	case CONNECTING:
2721
		c4iw_put_ep(&ep->parent_ep->com);
2722 2723
		break;
	case MPA_REQ_WAIT:
S
Steve Wise 已提交
2724
		(void)stop_ep_timer(ep);
2725 2726
		break;
	case MPA_REQ_SENT:
S
Steve Wise 已提交
2727
		(void)stop_ep_timer(ep);
2728
		if (mpa_rev == 1 || (mpa_rev == 2 && ep->tried_with_mpa_v1))
2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742
			connect_reply_upcall(ep, -ECONNRESET);
		else {
			/*
			 * we just don't send notification upwards because we
			 * want to retry with mpa_v1 without upper layers even
			 * knowing it.
			 *
			 * do some housekeeping so as to re-initiate the
			 * connection
			 */
			PDBG("%s: mpa_rev=%d. Retrying with mpav1\n", __func__,
			     mpa_rev);
			ep->retry_with_mpa_v1 = 1;
		}
2743 2744 2745 2746 2747 2748 2749
		break;
	case MPA_REP_SENT:
		break;
	case MPA_REQ_RCVD:
		break;
	case MORIBUND:
	case CLOSING:
2750
		stop_ep_timer(ep);
2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768
		/*FALLTHROUGH*/
	case FPDU_MODE:
		if (ep->com.cm_id && ep->com.qp) {
			attrs.next_state = C4IW_QP_STATE_ERROR;
			ret = c4iw_modify_qp(ep->com.qp->rhp,
				     ep->com.qp, C4IW_QP_ATTR_NEXT_STATE,
				     &attrs, 1);
			if (ret)
				printk(KERN_ERR MOD
				       "%s - qp <- error failed!\n",
				       __func__);
		}
		peer_abort_upcall(ep);
		break;
	case ABORTING:
		break;
	case DEAD:
		PDBG("%s PEER_ABORT IN DEAD STATE!!!!\n", __func__);
2769
		mutex_unlock(&ep->com.mutex);
2770 2771 2772 2773 2774 2775 2776 2777
		return 0;
	default:
		BUG_ON(1);
		break;
	}
	dst_confirm(ep->dst);
	if (ep->com.state != ABORTING) {
		__state_set(&ep->com, DEAD);
2778 2779 2780
		/* we don't release if we want to retry with mpa_v1 */
		if (!ep->retry_with_mpa_v1)
			release = 1;
2781
	}
2782
	mutex_unlock(&ep->com.mutex);
2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799

	rpl_skb = get_skb(skb, sizeof(*rpl), GFP_KERNEL);
	if (!rpl_skb) {
		printk(KERN_ERR MOD "%s - cannot allocate skb!\n",
		       __func__);
		release = 1;
		goto out;
	}
	set_wr_txq(skb, CPL_PRIORITY_DATA, ep->txq_idx);
	rpl = (struct cpl_abort_rpl *) skb_put(rpl_skb, sizeof(*rpl));
	INIT_TP_WR(rpl, ep->hwtid);
	OPCODE_TID(rpl) = cpu_to_be32(MK_OPCODE_TID(CPL_ABORT_RPL, ep->hwtid));
	rpl->cmd = CPL_ABORT_NO_RST;
	c4iw_ofld_send(&ep->com.dev->rdev, rpl_skb);
out:
	if (release)
		release_ep_resources(ep);
2800
	else if (ep->retry_with_mpa_v1) {
H
Hariprasad S 已提交
2801 2802 2803
		if (ep->com.remote_addr.ss_family == AF_INET6) {
			struct sockaddr_in6 *sin6 =
					(struct sockaddr_in6 *)
2804
					&ep->com.local_addr;
H
Hariprasad S 已提交
2805 2806 2807 2808 2809
			cxgb4_clip_release(
					ep->com.dev->rdev.lldi.ports[0],
					(const u32 *)&sin6->sin6_addr.s6_addr,
					1);
		}
2810
		remove_handle(ep->com.dev, &ep->com.dev->hwtid_idr, ep->hwtid);
2811 2812 2813 2814 2815 2816
		cxgb4_remove_tid(ep->com.dev->rdev.lldi.tids, 0, ep->hwtid);
		dst_release(ep->dst);
		cxgb4_l2t_release(ep->l2t);
		c4iw_reconnect(ep);
	}

2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834
	return 0;
}

static int close_con_rpl(struct c4iw_dev *dev, struct sk_buff *skb)
{
	struct c4iw_ep *ep;
	struct c4iw_qp_attributes attrs;
	struct cpl_close_con_rpl *rpl = cplhdr(skb);
	int release = 0;
	struct tid_info *t = dev->rdev.lldi.tids;
	unsigned int tid = GET_TID(rpl);

	ep = lookup_tid(t, tid);

	PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid);
	BUG_ON(!ep);

	/* The cm_id may be null if we failed to connect */
2835
	mutex_lock(&ep->com.mutex);
2836
	set_bit(CLOSE_CON_RPL, &ep->com.history);
2837 2838 2839 2840 2841
	switch (ep->com.state) {
	case CLOSING:
		__state_set(&ep->com, MORIBUND);
		break;
	case MORIBUND:
S
Steve Wise 已提交
2842
		(void)stop_ep_timer(ep);
2843 2844 2845 2846 2847 2848 2849
		if ((ep->com.cm_id) && (ep->com.qp)) {
			attrs.next_state = C4IW_QP_STATE_IDLE;
			c4iw_modify_qp(ep->com.qp->rhp,
					     ep->com.qp,
					     C4IW_QP_ATTR_NEXT_STATE,
					     &attrs, 1);
		}
2850
		close_complete_upcall(ep, 0);
2851 2852 2853 2854 2855 2856 2857 2858 2859 2860
		__state_set(&ep->com, DEAD);
		release = 1;
		break;
	case ABORTING:
	case DEAD:
		break;
	default:
		BUG_ON(1);
		break;
	}
2861
	mutex_unlock(&ep->com.mutex);
2862 2863 2864 2865 2866 2867 2868
	if (release)
		release_ep_resources(ep);
	return 0;
}

static int terminate(struct c4iw_dev *dev, struct sk_buff *skb)
{
2869
	struct cpl_rdma_terminate *rpl = cplhdr(skb);
2870
	struct tid_info *t = dev->rdev.lldi.tids;
2871 2872 2873
	unsigned int tid = GET_TID(rpl);
	struct c4iw_ep *ep;
	struct c4iw_qp_attributes attrs;
2874 2875

	ep = lookup_tid(t, tid);
2876
	BUG_ON(!ep);
2877

2878
	if (ep && ep->com.qp) {
2879 2880 2881 2882 2883 2884
		printk(KERN_WARNING MOD "TERM received tid %u qpid %u\n", tid,
		       ep->com.qp->wq.sq.qid);
		attrs.next_state = C4IW_QP_STATE_TERMINATE;
		c4iw_modify_qp(ep->com.qp->rhp, ep->com.qp,
			       C4IW_QP_ATTR_NEXT_STATE, &attrs, 1);
	} else
2885
		printk(KERN_WARNING MOD "TERM received tid %u no ep/qp\n", tid);
2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906

	return 0;
}

/*
 * Upcall from the adapter indicating data has been transmitted.
 * For us its just the single MPA request or reply.  We can now free
 * the skb holding the mpa message.
 */
static int fw4_ack(struct c4iw_dev *dev, struct sk_buff *skb)
{
	struct c4iw_ep *ep;
	struct cpl_fw4_ack *hdr = cplhdr(skb);
	u8 credits = hdr->credits;
	unsigned int tid = GET_TID(hdr);
	struct tid_info *t = dev->rdev.lldi.tids;


	ep = lookup_tid(t, tid);
	PDBG("%s ep %p tid %u credits %u\n", __func__, ep, ep->hwtid, credits);
	if (credits == 0) {
2907 2908
		PDBG("%s 0 credit ack ep %p tid %u state %u\n",
		     __func__, ep, ep->hwtid, state_read(&ep->com));
2909 2910 2911 2912 2913 2914 2915 2916 2917 2918
		return 0;
	}

	dst_confirm(ep->dst);
	if (ep->mpa_skb) {
		PDBG("%s last streaming msg ack ep %p tid %u state %u "
		     "initiator %u freeing skb\n", __func__, ep, ep->hwtid,
		     state_read(&ep->com), ep->mpa_attr.initiator ? 1 : 0);
		kfree_skb(ep->mpa_skb);
		ep->mpa_skb = NULL;
2919 2920 2921 2922
		mutex_lock(&ep->com.mutex);
		if (test_bit(STOP_MPA_TIMER, &ep->com.flags))
			stop_ep_timer(ep);
		mutex_unlock(&ep->com.mutex);
2923 2924 2925 2926 2927 2928
	}
	return 0;
}

int c4iw_reject_cr(struct iw_cm_id *cm_id, const void *pdata, u8 pdata_len)
{
2929 2930
	int err = 0;
	int disconnect = 0;
2931 2932 2933
	struct c4iw_ep *ep = to_ep(cm_id);
	PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid);

2934 2935 2936
	mutex_lock(&ep->com.mutex);
	if (ep->com.state == DEAD) {
		mutex_unlock(&ep->com.mutex);
2937 2938 2939
		c4iw_put_ep(&ep->com);
		return -ECONNRESET;
	}
2940
	set_bit(ULP_REJECT, &ep->com.history);
2941
	BUG_ON(ep->com.state != MPA_REQ_RCVD);
2942
	if (mpa_rev == 0)
2943
		disconnect = 2;
2944 2945
	else {
		err = send_mpa_reject(ep, pdata, pdata_len);
2946
		disconnect = 1;
2947
	}
2948
	mutex_unlock(&ep->com.mutex);
2949 2950
	if (disconnect) {
		stop_ep_timer(ep);
2951
		err = c4iw_ep_disconnect(ep, disconnect == 2, GFP_KERNEL);
2952
	}
2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964
	c4iw_put_ep(&ep->com);
	return 0;
}

int c4iw_accept_cr(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param)
{
	int err;
	struct c4iw_qp_attributes attrs;
	enum c4iw_qp_attr_mask mask;
	struct c4iw_ep *ep = to_ep(cm_id);
	struct c4iw_dev *h = to_c4iw_dev(cm_id->device);
	struct c4iw_qp *qp = get_qhp(h, conn_param->qpn);
2965
	int abort = 0;
2966 2967

	PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid);
2968 2969 2970

	mutex_lock(&ep->com.mutex);
	if (ep->com.state == DEAD) {
2971
		err = -ECONNRESET;
2972
		goto err_out;
2973 2974
	}

2975
	BUG_ON(ep->com.state != MPA_REQ_RCVD);
2976 2977
	BUG_ON(!qp);

2978
	set_bit(ULP_ACCEPT, &ep->com.history);
2979 2980
	if ((conn_param->ord > cur_max_read_depth(ep->com.dev)) ||
	    (conn_param->ird > cur_max_read_depth(ep->com.dev))) {
2981
		err = -EINVAL;
2982
		goto err_abort;
2983 2984
	}

2985 2986
	if (ep->mpa_attr.version == 2 && ep->mpa_attr.enhanced_rdma_conn) {
		if (conn_param->ord > ep->ird) {
2987 2988 2989 2990 2991 2992 2993 2994
			if (RELAXED_IRD_NEGOTIATION) {
				ep->ord = ep->ird;
			} else {
				ep->ird = conn_param->ird;
				ep->ord = conn_param->ord;
				send_mpa_reject(ep, conn_param->private_data,
						conn_param->private_data_len);
				err = -ENOMEM;
2995
				goto err_abort;
2996
			}
2997
		}
2998 2999 3000 3001 3002
		if (conn_param->ird < ep->ord) {
			if (RELAXED_IRD_NEGOTIATION &&
			    ep->ord <= h->rdev.lldi.max_ordird_qp) {
				conn_param->ird = ep->ord;
			} else {
3003
				err = -ENOMEM;
3004
				goto err_abort;
3005 3006 3007
			}
		}
	}
3008 3009 3010
	ep->ird = conn_param->ird;
	ep->ord = conn_param->ord;

3011
	if (ep->mpa_attr.version == 1) {
3012 3013
		if (peer2peer && ep->ird == 0)
			ep->ird = 1;
3014 3015 3016
	} else {
		if (peer2peer &&
		    (ep->mpa_attr.p2p_type != FW_RI_INIT_P2PTYPE_DISABLED) &&
3017
		    (p2p_type == FW_RI_INIT_P2PTYPE_READ_REQ) && ep->ird == 0)
3018 3019
			ep->ird = 1;
	}
3020 3021 3022

	PDBG("%s %d ird %d ord %d\n", __func__, __LINE__, ep->ird, ep->ord);

3023
	ep->com.cm_id = cm_id;
3024
	ref_cm_id(&ep->com);
3025
	ep->com.qp = qp;
3026
	ref_qp(ep);
3027

3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044
	/* bind QP to EP and move to RTS */
	attrs.mpa_attr = ep->mpa_attr;
	attrs.max_ird = ep->ird;
	attrs.max_ord = ep->ord;
	attrs.llp_stream_handle = ep;
	attrs.next_state = C4IW_QP_STATE_RTS;

	/* bind QP and TID with INIT_WR */
	mask = C4IW_QP_ATTR_NEXT_STATE |
			     C4IW_QP_ATTR_LLP_STREAM_HANDLE |
			     C4IW_QP_ATTR_MPA_ATTR |
			     C4IW_QP_ATTR_MAX_IRD |
			     C4IW_QP_ATTR_MAX_ORD;

	err = c4iw_modify_qp(ep->com.qp->rhp,
			     ep->com.qp, mask, &attrs, 1);
	if (err)
3045
		goto err_deref_cm_id;
3046 3047

	set_bit(STOP_MPA_TIMER, &ep->com.flags);
3048 3049 3050
	err = send_mpa_reply(ep, conn_param->private_data,
			     conn_param->private_data_len);
	if (err)
3051
		goto err_deref_cm_id;
3052

3053
	__state_set(&ep->com, FPDU_MODE);
3054
	established_upcall(ep);
3055
	mutex_unlock(&ep->com.mutex);
3056 3057
	c4iw_put_ep(&ep->com);
	return 0;
3058
err_deref_cm_id:
3059
	deref_cm_id(&ep->com);
3060 3061 3062
err_abort:
	abort = 1;
err_out:
3063
	mutex_unlock(&ep->com.mutex);
3064 3065
	if (abort)
		c4iw_ep_disconnect(ep, 1, GFP_KERNEL);
3066 3067 3068 3069
	c4iw_put_ep(&ep->com);
	return err;
}

3070 3071 3072 3073
static int pick_local_ipaddrs(struct c4iw_dev *dev, struct iw_cm_id *cm_id)
{
	struct in_device *ind;
	int found = 0;
3074 3075
	struct sockaddr_in *laddr = (struct sockaddr_in *)&cm_id->m_local_addr;
	struct sockaddr_in *raddr = (struct sockaddr_in *)&cm_id->m_remote_addr;
3076 3077 3078 3079 3080 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118 3119

	ind = in_dev_get(dev->rdev.lldi.ports[0]);
	if (!ind)
		return -EADDRNOTAVAIL;
	for_primary_ifa(ind) {
		laddr->sin_addr.s_addr = ifa->ifa_address;
		raddr->sin_addr.s_addr = ifa->ifa_address;
		found = 1;
		break;
	}
	endfor_ifa(ind);
	in_dev_put(ind);
	return found ? 0 : -EADDRNOTAVAIL;
}

static int get_lladdr(struct net_device *dev, struct in6_addr *addr,
		      unsigned char banned_flags)
{
	struct inet6_dev *idev;
	int err = -EADDRNOTAVAIL;

	rcu_read_lock();
	idev = __in6_dev_get(dev);
	if (idev != NULL) {
		struct inet6_ifaddr *ifp;

		read_lock_bh(&idev->lock);
		list_for_each_entry(ifp, &idev->addr_list, if_list) {
			if (ifp->scope == IFA_LINK &&
			    !(ifp->flags & banned_flags)) {
				memcpy(addr, &ifp->addr, 16);
				err = 0;
				break;
			}
		}
		read_unlock_bh(&idev->lock);
	}
	rcu_read_unlock();
	return err;
}

static int pick_local_ip6addrs(struct c4iw_dev *dev, struct iw_cm_id *cm_id)
{
	struct in6_addr uninitialized_var(addr);
3120 3121
	struct sockaddr_in6 *la6 = (struct sockaddr_in6 *)&cm_id->m_local_addr;
	struct sockaddr_in6 *ra6 = (struct sockaddr_in6 *)&cm_id->m_remote_addr;
3122

3123
	if (!get_lladdr(dev->rdev.lldi.ports[0], &addr, IFA_F_TENTATIVE)) {
3124 3125 3126 3127 3128 3129 3130
		memcpy(la6->sin6_addr.s6_addr, &addr, 16);
		memcpy(ra6->sin6_addr.s6_addr, &addr, 16);
		return 0;
	}
	return -EADDRNOTAVAIL;
}

3131 3132 3133 3134
int c4iw_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param)
{
	struct c4iw_dev *dev = to_c4iw_dev(cm_id->device);
	struct c4iw_ep *ep;
3135
	int err = 0;
3136 3137 3138 3139
	struct sockaddr_in *laddr;
	struct sockaddr_in *raddr;
	struct sockaddr_in6 *laddr6;
	struct sockaddr_in6 *raddr6;
3140 3141
	__u8 *ra;
	int iptype;
3142

3143 3144
	if ((conn_param->ord > cur_max_read_depth(dev)) ||
	    (conn_param->ird > cur_max_read_depth(dev))) {
3145 3146 3147
		err = -EINVAL;
		goto out;
	}
3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165
	ep = alloc_ep(sizeof(*ep), GFP_KERNEL);
	if (!ep) {
		printk(KERN_ERR MOD "%s - cannot alloc ep.\n", __func__);
		err = -ENOMEM;
		goto out;
	}
	init_timer(&ep->timer);
	ep->plen = conn_param->private_data_len;
	if (ep->plen)
		memcpy(ep->mpa_pkt + sizeof(struct mpa_message),
		       conn_param->private_data, ep->plen);
	ep->ird = conn_param->ird;
	ep->ord = conn_param->ord;

	if (peer2peer && ep->ord == 0)
		ep->ord = 1;

	ep->com.cm_id = cm_id;
3166 3167
	ref_cm_id(&ep->com);
	ep->com.dev = dev;
3168
	ep->com.qp = get_qhp(dev, conn_param->qpn);
3169 3170 3171
	if (!ep->com.qp) {
		PDBG("%s qpn 0x%x not found!\n", __func__, conn_param->qpn);
		err = -EINVAL;
3172
		goto fail1;
3173
	}
3174
	ref_qp(ep);
3175 3176 3177 3178 3179 3180 3181 3182 3183 3184
	PDBG("%s qpn 0x%x qp %p cm_id %p\n", __func__, conn_param->qpn,
	     ep->com.qp, cm_id);

	/*
	 * Allocate an active TID to initiate a TCP connection.
	 */
	ep->atid = cxgb4_alloc_atid(dev->rdev.lldi.tids, ep);
	if (ep->atid == -1) {
		printk(KERN_ERR MOD "%s - cannot alloc atid.\n", __func__);
		err = -ENOMEM;
3185
		goto fail1;
3186
	}
3187
	insert_handle(dev, &dev->atid_idr, ep, ep->atid);
3188

3189
	memcpy(&ep->com.local_addr, &cm_id->m_local_addr,
3190
	       sizeof(ep->com.local_addr));
3191
	memcpy(&ep->com.remote_addr, &cm_id->m_remote_addr,
3192 3193
	       sizeof(ep->com.remote_addr));

3194 3195 3196 3197
	laddr = (struct sockaddr_in *)&ep->com.local_addr;
	raddr = (struct sockaddr_in *)&ep->com.remote_addr;
	laddr6 = (struct sockaddr_in6 *)&ep->com.local_addr;
	raddr6 = (struct sockaddr_in6 *) &ep->com.remote_addr;
3198

3199
	if (cm_id->m_remote_addr.ss_family == AF_INET) {
3200 3201
		iptype = 4;
		ra = (__u8 *)&raddr->sin_addr;
3202

3203 3204 3205 3206 3207 3208
		/*
		 * Handle loopback requests to INADDR_ANY.
		 */
		if ((__force int)raddr->sin_addr.s_addr == INADDR_ANY) {
			err = pick_local_ipaddrs(dev, cm_id);
			if (err)
3209
				goto fail1;
3210 3211 3212 3213 3214 3215 3216 3217
		}

		/* find a route */
		PDBG("%s saddr %pI4 sport 0x%x raddr %pI4 rport 0x%x\n",
		     __func__, &laddr->sin_addr, ntohs(laddr->sin_port),
		     ra, ntohs(raddr->sin_port));
		ep->dst = find_route(dev, laddr->sin_addr.s_addr,
				     raddr->sin_addr.s_addr, laddr->sin_port,
H
Hariprasad S 已提交
3218
				     raddr->sin_port, cm_id->tos);
3219 3220 3221 3222 3223 3224 3225 3226 3227 3228
	} else {
		iptype = 6;
		ra = (__u8 *)&raddr6->sin6_addr;

		/*
		 * Handle loopback requests to INADDR_ANY.
		 */
		if (ipv6_addr_type(&raddr6->sin6_addr) == IPV6_ADDR_ANY) {
			err = pick_local_ip6addrs(dev, cm_id);
			if (err)
3229
				goto fail1;
3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242
		}

		/* find a route */
		PDBG("%s saddr %pI6 sport 0x%x raddr %pI6 rport 0x%x\n",
		     __func__, laddr6->sin6_addr.s6_addr,
		     ntohs(laddr6->sin6_port),
		     raddr6->sin6_addr.s6_addr, ntohs(raddr6->sin6_port));
		ep->dst = find_route6(dev, laddr6->sin6_addr.s6_addr,
				      raddr6->sin6_addr.s6_addr,
				      laddr6->sin6_port, raddr6->sin6_port, 0,
				      raddr6->sin6_scope_id);
	}
	if (!ep->dst) {
3243 3244
		printk(KERN_ERR MOD "%s - cannot find route.\n", __func__);
		err = -EHOSTUNREACH;
3245
		goto fail2;
3246 3247
	}

3248
	err = import_ep(ep, iptype, ra, ep->dst, ep->com.dev, true,
H
Hariprasad S 已提交
3249
			ep->com.dev->rdev.lldi.adapter_type, cm_id->tos);
3250
	if (err) {
3251
		printk(KERN_ERR MOD "%s - cannot alloc l2e.\n", __func__);
3252
		goto fail3;
3253 3254 3255 3256 3257 3258 3259
	}

	PDBG("%s txq_idx %u tx_chan %u smac_idx %u rss_qid %u l2t_idx %u\n",
		__func__, ep->txq_idx, ep->tx_chan, ep->smac_idx, ep->rss_qid,
		ep->l2t->idx);

	state_set(&ep->com, CONNECTING);
H
Hariprasad S 已提交
3260
	ep->tos = cm_id->tos;
3261 3262 3263 3264 3265 3266 3267 3268

	/* send connect request to rnic */
	err = send_connect(ep);
	if (!err)
		goto out;

	cxgb4_l2t_release(ep->l2t);
fail3:
3269 3270
	dst_release(ep->dst);
fail2:
3271
	remove_handle(ep->com.dev, &ep->com.dev->atid_idr, ep->atid);
3272
	cxgb4_free_atid(ep->com.dev->rdev.lldi.tids, ep->atid);
3273
fail1:
3274
	deref_cm_id(&ep->com);
3275 3276 3277 3278 3279
	c4iw_put_ep(&ep->com);
out:
	return err;
}

3280 3281 3282
static int create_server6(struct c4iw_dev *dev, struct c4iw_listen_ep *ep)
{
	int err;
3283
	struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)
3284
				    &ep->com.local_addr;
3285

3286 3287 3288 3289 3290 3291
	if (ipv6_addr_type(&sin6->sin6_addr) != IPV6_ADDR_ANY) {
		err = cxgb4_clip_get(ep->com.dev->rdev.lldi.ports[0],
				     (const u32 *)&sin6->sin6_addr.s6_addr, 1);
		if (err)
			return err;
	}
3292 3293 3294 3295 3296 3297 3298 3299 3300
	c4iw_init_wr_wait(&ep->com.wr_wait);
	err = cxgb4_create_server6(ep->com.dev->rdev.lldi.ports[0],
				   ep->stid, &sin6->sin6_addr,
				   sin6->sin6_port,
				   ep->com.dev->rdev.lldi.rxq_ids[0]);
	if (!err)
		err = c4iw_wait_for_reply(&ep->com.dev->rdev,
					  &ep->com.wr_wait,
					  0, 0, __func__);
3301 3302
	else if (err > 0)
		err = net_xmit_errno(err);
3303 3304 3305
	if (err) {
		cxgb4_clip_release(ep->com.dev->rdev.lldi.ports[0],
				   (const u32 *)&sin6->sin6_addr.s6_addr, 1);
3306 3307 3308
		pr_err("cxgb4_create_server6/filter failed err %d stid %d laddr %pI6 lport %d\n",
		       err, ep->stid,
		       sin6->sin6_addr.s6_addr, ntohs(sin6->sin6_port));
3309
	}
3310 3311 3312 3313 3314 3315
	return err;
}

static int create_server4(struct c4iw_dev *dev, struct c4iw_listen_ep *ep)
{
	int err;
3316
	struct sockaddr_in *sin = (struct sockaddr_in *)
3317
				  &ep->com.local_addr;
3318 3319 3320 3321 3322 3323 3324 3325

	if (dev->rdev.lldi.enable_fw_ofld_conn) {
		do {
			err = cxgb4_create_server_filter(
				ep->com.dev->rdev.lldi.ports[0], ep->stid,
				sin->sin_addr.s_addr, sin->sin_port, 0,
				ep->com.dev->rdev.lldi.rxq_ids[0], 0, 0);
			if (err == -EBUSY) {
3326 3327 3328 3329
				if (c4iw_fatal_error(&ep->com.dev->rdev)) {
					err = -EIO;
					break;
				}
3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342
				set_current_state(TASK_UNINTERRUPTIBLE);
				schedule_timeout(usecs_to_jiffies(100));
			}
		} while (err == -EBUSY);
	} else {
		c4iw_init_wr_wait(&ep->com.wr_wait);
		err = cxgb4_create_server(ep->com.dev->rdev.lldi.ports[0],
				ep->stid, sin->sin_addr.s_addr, sin->sin_port,
				0, ep->com.dev->rdev.lldi.rxq_ids[0]);
		if (!err)
			err = c4iw_wait_for_reply(&ep->com.dev->rdev,
						  &ep->com.wr_wait,
						  0, 0, __func__);
3343 3344
		else if (err > 0)
			err = net_xmit_errno(err);
3345 3346 3347 3348 3349 3350 3351 3352
	}
	if (err)
		pr_err("cxgb4_create_server/filter failed err %d stid %d laddr %pI4 lport %d\n"
		       , err, ep->stid,
		       &sin->sin_addr, ntohs(sin->sin_port));
	return err;
}

3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368
int c4iw_create_listen(struct iw_cm_id *cm_id, int backlog)
{
	int err = 0;
	struct c4iw_dev *dev = to_c4iw_dev(cm_id->device);
	struct c4iw_listen_ep *ep;

	might_sleep();

	ep = alloc_ep(sizeof(*ep), GFP_KERNEL);
	if (!ep) {
		printk(KERN_ERR MOD "%s - cannot alloc ep.\n", __func__);
		err = -ENOMEM;
		goto fail1;
	}
	PDBG("%s ep %p\n", __func__, ep);
	ep->com.cm_id = cm_id;
3369
	ref_cm_id(&ep->com);
3370 3371
	ep->com.dev = dev;
	ep->backlog = backlog;
3372
	memcpy(&ep->com.local_addr, &cm_id->m_local_addr,
3373
	       sizeof(ep->com.local_addr));
3374 3375 3376 3377

	/*
	 * Allocate a server TID.
	 */
3378 3379
	if (dev->rdev.lldi.enable_fw_ofld_conn &&
	    ep->com.local_addr.ss_family == AF_INET)
3380
		ep->stid = cxgb4_alloc_sftid(dev->rdev.lldi.tids,
3381
					     cm_id->m_local_addr.ss_family, ep);
3382
	else
3383
		ep->stid = cxgb4_alloc_stid(dev->rdev.lldi.tids,
3384
					    cm_id->m_local_addr.ss_family, ep);
3385

3386
	if (ep->stid == -1) {
3387
		printk(KERN_ERR MOD "%s - cannot alloc stid.\n", __func__);
3388 3389 3390
		err = -ENOMEM;
		goto fail2;
	}
3391
	insert_handle(dev, &dev->stid_idr, ep, ep->stid);
3392

3393 3394
	memcpy(&ep->com.local_addr, &cm_id->m_local_addr,
	       sizeof(ep->com.local_addr));
3395

3396
	state_set(&ep->com, LISTEN);
3397 3398 3399 3400
	if (ep->com.local_addr.ss_family == AF_INET)
		err = create_server4(dev, ep);
	else
		err = create_server6(dev, ep);
3401 3402 3403 3404
	if (!err) {
		cm_id->provider_data = ep;
		goto out;
	}
3405

3406 3407
	cxgb4_free_stid(ep->com.dev->rdev.lldi.tids, ep->stid,
			ep->com.local_addr.ss_family);
3408
fail2:
3409
	deref_cm_id(&ep->com);
3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422 3423 3424
	c4iw_put_ep(&ep->com);
fail1:
out:
	return err;
}

int c4iw_destroy_listen(struct iw_cm_id *cm_id)
{
	int err;
	struct c4iw_listen_ep *ep = to_listen_ep(cm_id);

	PDBG("%s ep %p\n", __func__, ep);

	might_sleep();
	state_set(&ep->com, DEAD);
3425 3426
	if (ep->com.dev->rdev.lldi.enable_fw_ofld_conn &&
	    ep->com.local_addr.ss_family == AF_INET) {
3427 3428 3429 3430
		err = cxgb4_remove_server_filter(
			ep->com.dev->rdev.lldi.ports[0], ep->stid,
			ep->com.dev->rdev.lldi.rxq_ids[0], 0);
	} else {
H
Hariprasad S 已提交
3431
		struct sockaddr_in6 *sin6;
3432
		c4iw_init_wr_wait(&ep->com.wr_wait);
3433 3434 3435
		err = cxgb4_remove_server(
				ep->com.dev->rdev.lldi.ports[0], ep->stid,
				ep->com.dev->rdev.lldi.rxq_ids[0], 0);
3436 3437 3438 3439
		if (err)
			goto done;
		err = c4iw_wait_for_reply(&ep->com.dev->rdev, &ep->com.wr_wait,
					  0, 0, __func__);
3440
		sin6 = (struct sockaddr_in6 *)&ep->com.local_addr;
H
Hariprasad S 已提交
3441 3442
		cxgb4_clip_release(ep->com.dev->rdev.lldi.ports[0],
				   (const u32 *)&sin6->sin6_addr.s6_addr, 1);
3443
	}
3444
	remove_handle(ep->com.dev, &ep->com.dev->stid_idr, ep->stid);
3445 3446
	cxgb4_free_stid(ep->com.dev->rdev.lldi.tids, ep->stid,
			ep->com.local_addr.ss_family);
3447
done:
3448
	deref_cm_id(&ep->com);
3449 3450 3451 3452 3453 3454 3455 3456 3457 3458 3459
	c4iw_put_ep(&ep->com);
	return err;
}

int c4iw_ep_disconnect(struct c4iw_ep *ep, int abrupt, gfp_t gfp)
{
	int ret = 0;
	int close = 0;
	int fatal = 0;
	struct c4iw_rdev *rdev;

3460
	mutex_lock(&ep->com.mutex);
3461 3462 3463 3464

	PDBG("%s ep %p state %s, abrupt %d\n", __func__, ep,
	     states[ep->com.state], abrupt);

3465 3466 3467 3468 3469 3470
	/*
	 * Ref the ep here in case we have fatal errors causing the
	 * ep to be released and freed.
	 */
	c4iw_get_ep(&ep->com);

3471 3472 3473
	rdev = &ep->com.dev->rdev;
	if (c4iw_fatal_error(rdev)) {
		fatal = 1;
3474
		close_complete_upcall(ep, -EIO);
3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487
		ep->com.state = DEAD;
	}
	switch (ep->com.state) {
	case MPA_REQ_WAIT:
	case MPA_REQ_SENT:
	case MPA_REQ_RCVD:
	case MPA_REP_SENT:
	case FPDU_MODE:
		close = 1;
		if (abrupt)
			ep->com.state = ABORTING;
		else {
			ep->com.state = CLOSING;
3488
			start_ep_timer(ep);
3489 3490 3491 3492 3493 3494 3495
		}
		set_bit(CLOSE_SENT, &ep->com.flags);
		break;
	case CLOSING:
		if (!test_and_set_bit(CLOSE_SENT, &ep->com.flags)) {
			close = 1;
			if (abrupt) {
S
Steve Wise 已提交
3496
				(void)stop_ep_timer(ep);
3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513
				ep->com.state = ABORTING;
			} else
				ep->com.state = MORIBUND;
		}
		break;
	case MORIBUND:
	case ABORTING:
	case DEAD:
		PDBG("%s ignoring disconnect ep %p state %u\n",
		     __func__, ep, ep->com.state);
		break;
	default:
		BUG();
		break;
	}

	if (close) {
S
Steve Wise 已提交
3514
		if (abrupt) {
3515
			set_bit(EP_DISC_ABORT, &ep->com.history);
3516
			close_complete_upcall(ep, -ECONNRESET);
S
Steve Wise 已提交
3517
			ret = send_abort(ep, NULL, gfp);
3518 3519
		} else {
			set_bit(EP_DISC_CLOSE, &ep->com.history);
3520
			ret = send_halfclose(ep, gfp);
3521
		}
3522
		if (ret) {
3523
			set_bit(EP_DISC_FAIL, &ep->com.history);
3524 3525 3526 3527
			if (!abrupt) {
				stop_ep_timer(ep);
				close_complete_upcall(ep, -EIO);
			}
3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540
			if (ep->com.qp) {
				struct c4iw_qp_attributes attrs;

				attrs.next_state = C4IW_QP_STATE_ERROR;
				ret = c4iw_modify_qp(ep->com.qp->rhp,
						     ep->com.qp,
						     C4IW_QP_ATTR_NEXT_STATE,
						     &attrs, 1);
				if (ret)
					pr_err(MOD
					       "%s - qp <- error failed!\n",
					       __func__);
			}
3541
			fatal = 1;
3542
		}
3543
	}
S
Steve Wise 已提交
3544
	mutex_unlock(&ep->com.mutex);
3545
	c4iw_put_ep(&ep->com);
3546 3547 3548 3549 3550
	if (fatal)
		release_ep_resources(ep);
	return ret;
}

3551 3552 3553 3554
static void active_ofld_conn_reply(struct c4iw_dev *dev, struct sk_buff *skb,
			struct cpl_fw6_msg_ofld_connection_wr_rpl *req)
{
	struct c4iw_ep *ep;
3555
	int atid = be32_to_cpu(req->tid);
3556

3557 3558
	ep = (struct c4iw_ep *)lookup_atid(dev->rdev.lldi.tids,
					   (__force u32) req->tid);
3559 3560 3561 3562 3563
	if (!ep)
		return;

	switch (req->retval) {
	case FW_ENOMEM:
3564 3565 3566 3567 3568
		set_bit(ACT_RETRY_NOMEM, &ep->com.history);
		if (ep->retry_count++ < ACT_OPEN_RETRY_COUNT) {
			send_fw_act_open_req(ep, atid);
			return;
		}
3569
	case FW_EADDRINUSE:
3570 3571 3572 3573 3574
		set_bit(ACT_RETRY_INUSE, &ep->com.history);
		if (ep->retry_count++ < ACT_OPEN_RETRY_COUNT) {
			send_fw_act_open_req(ep, atid);
			return;
		}
3575 3576 3577 3578 3579 3580
		break;
	default:
		pr_info("%s unexpected ofld conn wr retval %d\n",
		       __func__, req->retval);
		break;
	}
3581 3582 3583 3584 3585
	pr_err("active ofld_connect_wr failure %d atid %d\n",
	       req->retval, atid);
	mutex_lock(&dev->rdev.stats.lock);
	dev->rdev.stats.act_ofld_conn_fails++;
	mutex_unlock(&dev->rdev.stats.lock);
3586
	connect_reply_upcall(ep, status2errno(req->retval));
3587
	state_set(&ep->com, DEAD);
H
Hariprasad S 已提交
3588 3589
	if (ep->com.remote_addr.ss_family == AF_INET6) {
		struct sockaddr_in6 *sin6 =
3590
			(struct sockaddr_in6 *)&ep->com.local_addr;
H
Hariprasad S 已提交
3591 3592 3593
		cxgb4_clip_release(ep->com.dev->rdev.lldi.ports[0],
				   (const u32 *)&sin6->sin6_addr.s6_addr, 1);
	}
3594 3595 3596 3597 3598
	remove_handle(dev, &dev->atid_idr, atid);
	cxgb4_free_atid(dev->rdev.lldi.tids, atid);
	dst_release(ep->dst);
	cxgb4_l2t_release(ep->l2t);
	c4iw_put_ep(&ep->com);
3599 3600 3601 3602 3603 3604 3605 3606 3607
}

static void passive_ofld_conn_reply(struct c4iw_dev *dev, struct sk_buff *skb,
			struct cpl_fw6_msg_ofld_connection_wr_rpl *req)
{
	struct sk_buff *rpl_skb;
	struct cpl_pass_accept_req *cpl;
	int ret;

3608
	rpl_skb = (struct sk_buff *)(unsigned long)req->cookie;
3609 3610 3611
	BUG_ON(!rpl_skb);
	if (req->retval) {
		PDBG("%s passive open failure %d\n", __func__, req->retval);
3612 3613 3614
		mutex_lock(&dev->rdev.stats.lock);
		dev->rdev.stats.pas_ofld_conn_fails++;
		mutex_unlock(&dev->rdev.stats.lock);
3615 3616 3617 3618
		kfree_skb(rpl_skb);
	} else {
		cpl = (struct cpl_pass_accept_req *)cplhdr(rpl_skb);
		OPCODE_TID(cpl) = htonl(MK_OPCODE_TID(CPL_PASS_ACCEPT_REQ,
3619 3620
					(__force u32) htonl(
					(__force u32) req->tid)));
3621 3622 3623 3624 3625 3626 3627 3628
		ret = pass_accept_req(dev, rpl_skb);
		if (!ret)
			kfree_skb(rpl_skb);
	}
	return;
}

static int deferred_fw6_msg(struct c4iw_dev *dev, struct sk_buff *skb)
3629 3630
{
	struct cpl_fw6_msg *rpl = cplhdr(skb);
3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657
	struct cpl_fw6_msg_ofld_connection_wr_rpl *req;

	switch (rpl->type) {
	case FW6_TYPE_CQE:
		c4iw_ev_dispatch(dev, (struct t4_cqe *)&rpl->data[0]);
		break;
	case FW6_TYPE_OFLD_CONNECTION_WR_RPL:
		req = (struct cpl_fw6_msg_ofld_connection_wr_rpl *)rpl->data;
		switch (req->t_state) {
		case TCP_SYN_SENT:
			active_ofld_conn_reply(dev, skb, req);
			break;
		case TCP_SYN_RECV:
			passive_ofld_conn_reply(dev, skb, req);
			break;
		default:
			pr_err("%s unexpected ofld conn wr state %d\n",
			       __func__, req->t_state);
			break;
		}
		break;
	}
	return 0;
}

static void build_cpl_pass_accept_req(struct sk_buff *skb, int stid , u8 tos)
{
3658 3659 3660 3661
	__be32 l2info;
	__be16 hdr_len, vlantag, len;
	u16 eth_hdr_len;
	int tcp_hdr_len, ip_hdr_len;
3662 3663 3664 3665
	u8 intf;
	struct cpl_rx_pkt *cpl = cplhdr(skb);
	struct cpl_pass_accept_req *req;
	struct tcp_options_received tmp_opt;
3666
	struct c4iw_dev *dev;
3667
	enum chip_type type;
3668

3669
	dev = *((struct c4iw_dev **) (skb->cb + sizeof(void *)));
3670
	/* Store values from cpl_rx_pkt in temporary location. */
3671 3672 3673 3674
	vlantag = cpl->vlan;
	len = cpl->len;
	l2info  = cpl->l2info;
	hdr_len = cpl->hdr_len;
3675 3676 3677 3678 3679 3680 3681 3682 3683 3684
	intf = cpl->iff;

	__skb_pull(skb, sizeof(*req) + sizeof(struct rss_header));

	/*
	 * We need to parse the TCP options from SYN packet.
	 * to generate cpl_pass_accept_req.
	 */
	memset(&tmp_opt, 0, sizeof(tmp_opt));
	tcp_clear_options(&tmp_opt);
C
Christoph Paasch 已提交
3685
	tcp_parse_options(skb, &tmp_opt, 0, NULL);
3686 3687 3688

	req = (struct cpl_pass_accept_req *)__skb_push(skb, sizeof(*req));
	memset(req, 0, sizeof(*req));
3689 3690
	req->l2info = cpu_to_be16(SYN_INTF_V(intf) |
			 SYN_MAC_IDX_V(RX_MACIDX_G(
3691
			 be32_to_cpu(l2info))) |
3692
			 SYN_XACT_MATCH_F);
3693 3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712
	type = dev->rdev.lldi.adapter_type;
	tcp_hdr_len = RX_TCPHDR_LEN_G(be16_to_cpu(hdr_len));
	ip_hdr_len = RX_IPHDR_LEN_G(be16_to_cpu(hdr_len));
	req->hdr_len =
		cpu_to_be32(SYN_RX_CHAN_V(RX_CHAN_G(be32_to_cpu(l2info))));
	if (CHELSIO_CHIP_VERSION(type) <= CHELSIO_T5) {
		eth_hdr_len = is_t4(type) ?
				RX_ETHHDR_LEN_G(be32_to_cpu(l2info)) :
				RX_T5_ETHHDR_LEN_G(be32_to_cpu(l2info));
		req->hdr_len |= cpu_to_be32(TCP_HDR_LEN_V(tcp_hdr_len) |
					    IP_HDR_LEN_V(ip_hdr_len) |
					    ETH_HDR_LEN_V(eth_hdr_len));
	} else { /* T6 and later */
		eth_hdr_len = RX_T6_ETHHDR_LEN_G(be32_to_cpu(l2info));
		req->hdr_len |= cpu_to_be32(T6_TCP_HDR_LEN_V(tcp_hdr_len) |
					    T6_IP_HDR_LEN_V(ip_hdr_len) |
					    T6_ETH_HDR_LEN_V(eth_hdr_len));
	}
	req->vlan = vlantag;
	req->len = len;
3713 3714
	req->tos_stid = cpu_to_be32(PASS_OPEN_TID_V(stid) |
				    PASS_OPEN_TOS_V(tos));
3715 3716 3717 3718 3719 3720 3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733
	req->tcpopt.mss = htons(tmp_opt.mss_clamp);
	if (tmp_opt.wscale_ok)
		req->tcpopt.wsf = tmp_opt.snd_wscale;
	req->tcpopt.tstamp = tmp_opt.saw_tstamp;
	if (tmp_opt.sack_ok)
		req->tcpopt.sack = 1;
	OPCODE_TID(req) = htonl(MK_OPCODE_TID(CPL_PASS_ACCEPT_REQ, 0));
	return;
}

static void send_fw_pass_open_req(struct c4iw_dev *dev, struct sk_buff *skb,
				  __be32 laddr, __be16 lport,
				  __be32 raddr, __be16 rport,
				  u32 rcv_isn, u32 filter, u16 window,
				  u32 rss_qid, u8 port_id)
{
	struct sk_buff *req_skb;
	struct fw_ofld_connection_wr *req;
	struct cpl_pass_accept_req *cpl = cplhdr(skb);
3734
	int ret;
3735 3736 3737 3738

	req_skb = alloc_skb(sizeof(struct fw_ofld_connection_wr), GFP_KERNEL);
	req = (struct fw_ofld_connection_wr *)__skb_put(req_skb, sizeof(*req));
	memset(req, 0, sizeof(*req));
3739
	req->op_compl = htonl(WR_OP_V(FW_OFLD_CONNECTION_WR) | FW_WR_COMPL_F);
3740
	req->len16_pkd = htonl(FW_WR_LEN16_V(DIV_ROUND_UP(sizeof(*req), 16)));
3741
	req->le.version_cpl = htonl(FW_OFLD_CONNECTION_WR_CPL_F);
3742
	req->le.filter = (__force __be32) filter;
3743 3744 3745 3746 3747 3748 3749
	req->le.lport = lport;
	req->le.pport = rport;
	req->le.u.ipv4.lip = laddr;
	req->le.u.ipv4.pip = raddr;
	req->tcb.rcv_nxt = htonl(rcv_isn + 1);
	req->tcb.rcv_adv = htons(window);
	req->tcb.t_state_to_astid =
3750 3751 3752
		 htonl(FW_OFLD_CONNECTION_WR_T_STATE_V(TCP_SYN_RECV) |
			FW_OFLD_CONNECTION_WR_RCV_SCALE_V(cpl->tcpopt.wsf) |
			FW_OFLD_CONNECTION_WR_ASTID_V(
3753
			PASS_OPEN_TID_G(ntohl(cpl->tos_stid))));
3754 3755 3756 3757 3758

	/*
	 * We store the qid in opt2 which will be used by the firmware
	 * to send us the wr response.
	 */
3759
	req->tcb.opt2 = htonl(RSS_QUEUE_V(rss_qid));
3760 3761 3762 3763 3764 3765 3766

	/*
	 * We initialize the MSS index in TCB to 0xF.
	 * So that when driver sends cpl_pass_accept_rpl
	 * TCB picks up the correct value. If this was 0
	 * TP will ignore any value > 0 for MSS index.
	 */
3767
	req->tcb.opt0 = cpu_to_be64(MSS_IDX_V(0xF));
H
Hariprasad S 已提交
3768
	req->cookie = (uintptr_t)skb;
3769 3770

	set_wr_txq(req_skb, CPL_PRIORITY_CONTROL, port_id);
3771 3772 3773 3774 3775 3776 3777
	ret = cxgb4_ofld_send(dev->rdev.lldi.ports[0], req_skb);
	if (ret < 0) {
		pr_err("%s - cxgb4_ofld_send error %d - dropping\n", __func__,
		       ret);
		kfree_skb(skb);
		kfree_skb(req_skb);
	}
3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795 3796 3797 3798 3799 3800 3801 3802 3803
}

/*
 * Handler for CPL_RX_PKT message. Need to handle cpl_rx_pkt
 * messages when a filter is being used instead of server to
 * redirect a syn packet. When packets hit filter they are redirected
 * to the offload queue and driver tries to establish the connection
 * using firmware work request.
 */
static int rx_pkt(struct c4iw_dev *dev, struct sk_buff *skb)
{
	int stid;
	unsigned int filter;
	struct ethhdr *eh = NULL;
	struct vlan_ethhdr *vlan_eh = NULL;
	struct iphdr *iph;
	struct tcphdr *tcph;
	struct rss_header *rss = (void *)skb->data;
	struct cpl_rx_pkt *cpl = (void *)skb->data;
	struct cpl_pass_accept_req *req = (void *)(rss + 1);
	struct l2t_entry *e;
	struct dst_entry *dst;
	struct c4iw_ep *lep;
	u16 window;
	struct port_info *pi;
	struct net_device *pdev;
3804
	u16 rss_qid, eth_hdr_len;
3805 3806 3807 3808 3809
	int step;
	u32 tx_chan;
	struct neighbour *neigh;

	/* Drop all non-SYN packets */
3810
	if (!(cpl->l2info & cpu_to_be32(RXF_SYN_F)))
3811 3812 3813 3814 3815 3816 3817 3818 3819 3820 3821 3822
		goto reject;

	/*
	 * Drop all packets which did not hit the filter.
	 * Unlikely to happen.
	 */
	if (!(rss->filter_hit && rss->filter_tid))
		goto reject;

	/*
	 * Calculate the server tid from filter hit index from cpl_rx_pkt.
	 */
3823
	stid = (__force int) cpu_to_be32((__force u32) rss->hash_val);
3824 3825 3826 3827 3828 3829 3830

	lep = (struct c4iw_ep *)lookup_stid(dev->rdev.lldi.tids, stid);
	if (!lep) {
		PDBG("%s connect request on invalid stid %d\n", __func__, stid);
		goto reject;
	}

3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844 3845 3846
	switch (CHELSIO_CHIP_VERSION(dev->rdev.lldi.adapter_type)) {
	case CHELSIO_T4:
		eth_hdr_len = RX_ETHHDR_LEN_G(be32_to_cpu(cpl->l2info));
		break;
	case CHELSIO_T5:
		eth_hdr_len = RX_T5_ETHHDR_LEN_G(be32_to_cpu(cpl->l2info));
		break;
	case CHELSIO_T6:
		eth_hdr_len = RX_T6_ETHHDR_LEN_G(be32_to_cpu(cpl->l2info));
		break;
	default:
		pr_err("T%d Chip is not supported\n",
		       CHELSIO_CHIP_VERSION(dev->rdev.lldi.adapter_type));
		goto reject;
	}

3847
	if (eth_hdr_len == ETH_HLEN) {
3848 3849 3850 3851 3852 3853 3854 3855 3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867
		eh = (struct ethhdr *)(req + 1);
		iph = (struct iphdr *)(eh + 1);
	} else {
		vlan_eh = (struct vlan_ethhdr *)(req + 1);
		iph = (struct iphdr *)(vlan_eh + 1);
		skb->vlan_tci = ntohs(cpl->vlan);
	}

	if (iph->version != 0x4)
		goto reject;

	tcph = (struct tcphdr *)(iph + 1);
	skb_set_network_header(skb, (void *)iph - (void *)rss);
	skb_set_transport_header(skb, (void *)tcph - (void *)rss);
	skb_get(skb);

	PDBG("%s lip 0x%x lport %u pip 0x%x pport %u tos %d\n", __func__,
	     ntohl(iph->daddr), ntohs(tcph->dest), ntohl(iph->saddr),
	     ntohs(tcph->source), iph->tos);

3868 3869 3870
	dst = find_route(dev, iph->daddr, iph->saddr, tcph->dest, tcph->source,
			 iph->tos);
	if (!dst) {
3871 3872 3873 3874 3875 3876
		pr_err("%s - failed to find dst entry!\n",
		       __func__);
		goto reject;
	}
	neigh = dst_neigh_lookup_skb(dst, skb);

3877 3878 3879 3880 3881 3882
	if (!neigh) {
		pr_err("%s - failed to allocate neigh!\n",
		       __func__);
		goto free_dst;
	}

3883 3884 3885 3886 3887 3888 3889 3890
	if (neigh->dev->flags & IFF_LOOPBACK) {
		pdev = ip_dev_find(&init_net, iph->daddr);
		e = cxgb4_l2t_get(dev->rdev.lldi.l2t, neigh,
				    pdev, 0);
		pi = (struct port_info *)netdev_priv(pdev);
		tx_chan = cxgb4_port_chan(pdev);
		dev_put(pdev);
	} else {
3891
		pdev = get_real_dev(neigh->dev);
3892
		e = cxgb4_l2t_get(dev->rdev.lldi.l2t, neigh,
3893 3894 3895
					pdev, 0);
		pi = (struct port_info *)netdev_priv(pdev);
		tx_chan = cxgb4_port_chan(pdev);
3896
	}
3897
	neigh_release(neigh);
3898 3899 3900 3901 3902 3903 3904 3905
	if (!e) {
		pr_err("%s - failed to allocate l2t entry!\n",
		       __func__);
		goto free_dst;
	}

	step = dev->rdev.lldi.nrxq / dev->rdev.lldi.nchan;
	rss_qid = dev->rdev.lldi.rxq_ids[pi->port_id * step];
3906
	window = (__force u16) htons((__force u16)tcph->window);
3907 3908

	/* Calcuate filter portion for LE region. */
3909 3910 3911
	filter = (__force unsigned int) cpu_to_be32(cxgb4_select_ntuple(
						    dev->rdev.lldi.ports[0],
						    e));
3912 3913 3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925

	/*
	 * Synthesize the cpl_pass_accept_req. We have everything except the
	 * TID. Once firmware sends a reply with TID we update the TID field
	 * in cpl and pass it through the regular cpl_pass_accept_req path.
	 */
	build_cpl_pass_accept_req(skb, stid, iph->tos);
	send_fw_pass_open_req(dev, skb, iph->daddr, tcph->dest, iph->saddr,
			      tcph->source, ntohl(tcph->seq), filter, window,
			      rss_qid, pi->port_id);
	cxgb4_l2t_release(e);
free_dst:
	dst_release(dst);
reject:
3926 3927 3928
	return 0;
}

3929 3930 3931 3932
/*
 * These are the real handlers that are called from a
 * work queue.
 */
3933
static c4iw_handler_func work_handlers[NUM_CPL_CMDS + NUM_FAKE_CPLS] = {
3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946
	[CPL_ACT_ESTABLISH] = act_establish,
	[CPL_ACT_OPEN_RPL] = act_open_rpl,
	[CPL_RX_DATA] = rx_data,
	[CPL_ABORT_RPL_RSS] = abort_rpl,
	[CPL_ABORT_RPL] = abort_rpl,
	[CPL_PASS_OPEN_RPL] = pass_open_rpl,
	[CPL_CLOSE_LISTSRV_RPL] = close_listsrv_rpl,
	[CPL_PASS_ACCEPT_REQ] = pass_accept_req,
	[CPL_PASS_ESTABLISH] = pass_establish,
	[CPL_PEER_CLOSE] = peer_close,
	[CPL_ABORT_REQ_RSS] = peer_abort,
	[CPL_CLOSE_CON_RPL] = close_con_rpl,
	[CPL_RDMA_TERMINATE] = terminate,
3947
	[CPL_FW4_ACK] = fw4_ack,
3948
	[CPL_FW6_MSG] = deferred_fw6_msg,
3949
	[CPL_RX_PKT] = rx_pkt,
3950 3951
	[FAKE_CPL_PUT_EP_SAFE] = _put_ep_safe,
	[FAKE_CPL_PASS_PUT_EP_SAFE] = _put_pass_ep_safe
3952 3953 3954 3955 3956 3957 3958
};

static void process_timeout(struct c4iw_ep *ep)
{
	struct c4iw_qp_attributes attrs;
	int abort = 1;

3959
	mutex_lock(&ep->com.mutex);
3960 3961
	PDBG("%s ep %p tid %u state %d\n", __func__, ep, ep->hwtid,
	     ep->com.state);
3962
	set_bit(TIMEDOUT, &ep->com.history);
3963 3964 3965 3966 3967 3968
	switch (ep->com.state) {
	case MPA_REQ_SENT:
		__state_set(&ep->com, ABORTING);
		connect_reply_upcall(ep, -ETIMEDOUT);
		break;
	case MPA_REQ_WAIT:
3969
	case MPA_REP_SENT:
3970 3971 3972 3973 3974 3975 3976 3977 3978 3979 3980
		__state_set(&ep->com, ABORTING);
		break;
	case CLOSING:
	case MORIBUND:
		if (ep->com.cm_id && ep->com.qp) {
			attrs.next_state = C4IW_QP_STATE_ERROR;
			c4iw_modify_qp(ep->com.qp->rhp,
				     ep->com.qp, C4IW_QP_ATTR_NEXT_STATE,
				     &attrs, 1);
		}
		__state_set(&ep->com, ABORTING);
3981
		close_complete_upcall(ep, -ETIMEDOUT);
3982
		break;
S
Steve Wise 已提交
3983 3984 3985 3986 3987 3988 3989 3990 3991 3992
	case ABORTING:
	case DEAD:

		/*
		 * These states are expected if the ep timed out at the same
		 * time as another thread was calling stop_ep_timer().
		 * So we silently do nothing for these states.
		 */
		abort = 0;
		break;
3993
	default:
J
Julia Lawall 已提交
3994
		WARN(1, "%s unexpected state ep %p tid %u state %u\n",
3995 3996 3997
			__func__, ep, ep->hwtid, ep->com.state);
		abort = 0;
	}
3998
	mutex_unlock(&ep->com.mutex);
3999 4000
	if (abort)
		c4iw_ep_disconnect(ep, 1, GFP_KERNEL);
4001 4002 4003 4004 4005 4006 4007 4008 4009 4010 4011 4012 4013
	c4iw_put_ep(&ep->com);
}

static void process_timedout_eps(void)
{
	struct c4iw_ep *ep;

	spin_lock_irq(&timeout_lock);
	while (!list_empty(&timeout_list)) {
		struct list_head *tmp;

		tmp = timeout_list.next;
		list_del(tmp);
S
Steve Wise 已提交
4014 4015
		tmp->next = NULL;
		tmp->prev = NULL;
4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027
		spin_unlock_irq(&timeout_lock);
		ep = list_entry(tmp, struct c4iw_ep, entry);
		process_timeout(ep);
		spin_lock_irq(&timeout_lock);
	}
	spin_unlock_irq(&timeout_lock);
}

static void process_work(struct work_struct *work)
{
	struct sk_buff *skb = NULL;
	struct c4iw_dev *dev;
4028
	struct cpl_act_establish *rpl;
4029 4030 4031
	unsigned int opcode;
	int ret;

S
Steve Wise 已提交
4032
	process_timedout_eps();
4033 4034 4035 4036 4037 4038 4039 4040 4041
	while ((skb = skb_dequeue(&rxq))) {
		rpl = cplhdr(skb);
		dev = *((struct c4iw_dev **) (skb->cb + sizeof(void *)));
		opcode = rpl->ot.opcode;

		BUG_ON(!work_handlers[opcode]);
		ret = work_handlers[opcode](dev, skb);
		if (!ret)
			kfree_skb(skb);
S
Steve Wise 已提交
4042
		process_timedout_eps();
4043 4044 4045 4046 4047 4048 4049 4050
	}
}

static DECLARE_WORK(skb_work, process_work);

static void ep_timeout(unsigned long arg)
{
	struct c4iw_ep *ep = (struct c4iw_ep *)arg;
4051
	int kickit = 0;
4052 4053

	spin_lock(&timeout_lock);
4054
	if (!test_and_set_bit(TIMEOUT, &ep->com.flags)) {
S
Steve Wise 已提交
4055 4056 4057 4058 4059 4060 4061
		/*
		 * Only insert if it is not already on the list.
		 */
		if (!ep->entry.next) {
			list_add_tail(&ep->entry, &timeout_list);
			kickit = 1;
		}
4062
	}
4063
	spin_unlock(&timeout_lock);
4064 4065
	if (kickit)
		queue_work(workq, &skb_work);
4066 4067
}

4068 4069 4070 4071 4072 4073 4074 4075 4076 4077 4078 4079 4080 4081 4082 4083 4084 4085 4086 4087 4088 4089 4090 4091 4092 4093 4094
/*
 * All the CM events are handled on a work queue to have a safe context.
 */
static int sched(struct c4iw_dev *dev, struct sk_buff *skb)
{

	/*
	 * Save dev in the skb->cb area.
	 */
	*((struct c4iw_dev **) (skb->cb + sizeof(void *))) = dev;

	/*
	 * Queue the skb and schedule the worker thread.
	 */
	skb_queue_tail(&rxq, skb);
	queue_work(workq, &skb_work);
	return 0;
}

static int set_tcb_rpl(struct c4iw_dev *dev, struct sk_buff *skb)
{
	struct cpl_set_tcb_rpl *rpl = cplhdr(skb);

	if (rpl->status != CPL_ERR_NONE) {
		printk(KERN_ERR MOD "Unexpected SET_TCB_RPL status %u "
		       "for tid %u\n", rpl->status, GET_TID(rpl));
	}
4095
	kfree_skb(skb);
4096 4097 4098
	return 0;
}

4099 4100 4101 4102 4103 4104 4105 4106 4107
static int fw6_msg(struct c4iw_dev *dev, struct sk_buff *skb)
{
	struct cpl_fw6_msg *rpl = cplhdr(skb);
	struct c4iw_wr_wait *wr_waitp;
	int ret;

	PDBG("%s type %u\n", __func__, rpl->type);

	switch (rpl->type) {
4108
	case FW6_TYPE_WR_RPL:
4109
		ret = (int)((be64_to_cpu(rpl->data[0]) >> 8) & 0xff);
4110
		wr_waitp = (struct c4iw_wr_wait *)(__force unsigned long) rpl->data[1];
4111
		PDBG("%s wr_waitp %p ret %u\n", __func__, wr_waitp, ret);
4112 4113
		if (wr_waitp)
			c4iw_wake_up(wr_waitp, ret ? -ret : 0);
4114
		kfree_skb(skb);
4115
		break;
4116 4117
	case FW6_TYPE_CQE:
	case FW6_TYPE_OFLD_CONNECTION_WR_RPL:
4118
		sched(dev, skb);
4119
		break;
4120 4121 4122
	default:
		printk(KERN_ERR MOD "%s unexpected fw6 msg type %u\n", __func__,
		       rpl->type);
4123
		kfree_skb(skb);
4124 4125 4126 4127 4128
		break;
	}
	return 0;
}

S
Steve Wise 已提交
4129 4130 4131 4132 4133 4134 4135 4136
static int peer_abort_intr(struct c4iw_dev *dev, struct sk_buff *skb)
{
	struct cpl_abort_req_rss *req = cplhdr(skb);
	struct c4iw_ep *ep;
	struct tid_info *t = dev->rdev.lldi.tids;
	unsigned int tid = GET_TID(req);

	ep = lookup_tid(t, tid);
4137 4138 4139 4140 4141 4142
	if (!ep) {
		printk(KERN_WARNING MOD
		       "Abort on non-existent endpoint, tid %d\n", tid);
		kfree_skb(skb);
		return 0;
	}
4143
	if (is_neg_adv(req->status)) {
4144 4145 4146 4147 4148
		PDBG("%s Negative advice on abort- tid %u status %d (%s)\n",
		     __func__, ep->hwtid, req->status,
		     neg_adv_str(req->status));
		ep->stats.abort_neg_adv++;
		dev->rdev.stats.neg_adv++;
S
Steve Wise 已提交
4149 4150 4151 4152 4153 4154 4155 4156
		kfree_skb(skb);
		return 0;
	}
	PDBG("%s ep %p tid %u state %u\n", __func__, ep, ep->hwtid,
	     ep->com.state);

	/*
	 * Wake up any threads in rdma_init() or rdma_fini().
4157 4158
	 * However, if we are on MPAv2 and want to retry with MPAv1
	 * then, don't wake up yet.
S
Steve Wise 已提交
4159
	 */
4160 4161 4162 4163 4164
	if (mpa_rev == 2 && !ep->tried_with_mpa_v1) {
		if (ep->com.state != MPA_REQ_SENT)
			c4iw_wake_up(&ep->com.wr_wait, -ECONNRESET);
	} else
		c4iw_wake_up(&ep->com.wr_wait, -ECONNRESET);
S
Steve Wise 已提交
4165 4166 4167 4168
	sched(dev, skb);
	return 0;
}

4169 4170 4171 4172 4173 4174 4175 4176 4177 4178 4179 4180 4181 4182 4183 4184
/*
 * Most upcalls from the T4 Core go to sched() to
 * schedule the processing on a work queue.
 */
c4iw_handler_func c4iw_handlers[NUM_CPL_CMDS] = {
	[CPL_ACT_ESTABLISH] = sched,
	[CPL_ACT_OPEN_RPL] = sched,
	[CPL_RX_DATA] = sched,
	[CPL_ABORT_RPL_RSS] = sched,
	[CPL_ABORT_RPL] = sched,
	[CPL_PASS_OPEN_RPL] = sched,
	[CPL_CLOSE_LISTSRV_RPL] = sched,
	[CPL_PASS_ACCEPT_REQ] = sched,
	[CPL_PASS_ESTABLISH] = sched,
	[CPL_PEER_CLOSE] = sched,
	[CPL_CLOSE_CON_RPL] = sched,
S
Steve Wise 已提交
4185
	[CPL_ABORT_REQ_RSS] = peer_abort_intr,
4186 4187 4188
	[CPL_RDMA_TERMINATE] = sched,
	[CPL_FW4_ACK] = sched,
	[CPL_SET_TCB_RPL] = set_tcb_rpl,
4189 4190
	[CPL_FW6_MSG] = fw6_msg,
	[CPL_RX_PKT] = sched
4191 4192
};

4193 4194
int __init c4iw_cm_init(void)
{
4195
	spin_lock_init(&timeout_lock);
4196 4197 4198 4199 4200 4201 4202 4203 4204
	skb_queue_head_init(&rxq);

	workq = create_singlethread_workqueue("iw_cxgb4");
	if (!workq)
		return -ENOMEM;

	return 0;
}

4205
void c4iw_cm_term(void)
4206
{
4207
	WARN_ON(!list_empty(&timeout_list));
4208 4209 4210
	flush_workqueue(workq);
	destroy_workqueue(workq);
}